anyway to share power bi dashboard in free account
- User
anyway to share power bi dashboard in free account
- Assistant
Unfortunately, in a free Power BI account, you cannot directly share dashboards with others. Sharing dashboards is a feature available only with a Power BI Pro license or higher. However, there are a few workarounds you can explore:
1. Export to PDF or PowerPoint
You can export your dashboard or report as a PDF or PowerPoint file and share it with others via email or a file-sharing platform. Here's how:
- Open the report or dashboard.
- Click on the File menu and select either Export to PDF or Export to PowerPoint.
- Share the exported file with your audience.
2. Share a Static Screenshot
You can take a screenshot of your dashboard and share it as an image. While it won't be interactive, it allows you to share the visuals and insights with others.
3. Publish to Web (Limited Option)
If your organization permits it, you can use the Publish to Web option. Be cautious with this method, as it makes your dashboard publicly available.
- Open your report in Power BI Service.
- Click File > Publish to Web.
- Copy the link generated and share it with your audience.
Note: This option should only be used for non-confidential or public-facing data, as anyone with the link can access the report, and it will be indexed by search engines.
4. Collaboration via Power BI Desktop File (.pbix)
You can share the raw .pbix file (Power BI Desktop file) of your report/dashboard with others. The recipients can open the file in Power BI Desktop on their own machine. This requires them to have data access and understand the Power BI structure.
5. Upgrade to Power BI Pro or Use Power BI Premium Per User
If none of the above options work and you need full sharing features, you may need to upgrade to Power BI Pro (required for both the creator and recipient to share directly) or Power BI Premium Per User, which is a more advanced option. Power BI Pro is not free, but it allows seamless dashboard sharing within your organization.
If your situation changes or your organization supports paid versions of Power BI, this could improve collaboration.
- User
Publish to Web (Limited Option)
how to do it
- Assistant
The "Publish to Web" option in Power BI lets you embed your reports or dashboards into a public-facing webpage or generate a public link that anyone can access. However, this option should only be used for public-facing, non-sensitive, and non-confidential data, as the dashboard/report becomes exposed to a wider audience and can even be indexed by search engines.
Here's how you can use the "Publish to Web" feature:
Steps to Publish to Web in Power BI
-
Go to Power BI Service
- Open your browser and go to Power BI Service (https://app.powerbi.com).
- Log in with your Power BI account (free or Pro).
-
Open the Report You Want to Publish
- Navigate to the desired workspace, and open the report you wish to share.
- Make sure your report is finalized, as edits won't affect the published web link. If you edit the report later, you would need to republish the link for those changes.
-
Use the "Publish to Web" Option
- On the top-right of the report, click the File menu (or sometimes it's under the "Share" button).
- Click Publish to Web.
Note: If this option is not available, it could be because:
- Your administrator has disabled the "Publish to Web" feature for security purposes.
- The dataset or report you're trying to publish doesn't allow this feature.
-
Confirm the Warning
- You'll receive a warning message about making the report public. As mentioned earlier, everything will be accessible to anyone with the link, and it will NOT be secure or private.
- Confirm that you're aware of this by proceeding with publishing.
-
Copy the Embed Code or Link
- Once confirmed, Power BI will generate a public link and an embed code.
- You can:
- Use the link to share with others via email or social media.
- Use the embed code to embed the report into a website or blog.
-
Share the Link Safely
- Copy the link and share it with your audience. Anyone with this link will now be able to view the report.
Important Considerations
- Accessibility: The "Publish to Web" link makes the report publicly accessible; anyone with the URL can view it (no account is required). Be cautious if confidentiality is important.
- Disable Publishing: You can stop sharing the report by revoking the "Publish to Web" setting. You can do this in the settings for the report in Power BI Service.
- Check Permissions: If the "Publish to Web" option is disabled for your account or organization, contact your Power BI administrator to enable it (if permissible under your organization’s policy).
- Data Security: Do not use this feature for sensitive or private data. It is suitable for publicly shareable reports, such as promotional or general statistics.
How to Stop Sharing ("Unpublish")
If you wish to stop sharing:
- Go to Power BI Service.
- Navigate to the report settings.
- Look for an option to manage "Published to Web" reports.
- Delete the public link to unpublish the report.
This method is one of the few sharing options available in a free Power BI account but should be used with caution due to its public nature.
-
- User
“Edit and refine this [ As checked the LogMeIn beta is only working mac devices with intel chips , it is not working in devices with apple silicon chips(m1 and m2 series). ] to ensure it meets high-quality standards. . Additionally, assess the use of active voice, appropriate word choice, and proper citation and referencing. Aim to enhance the audience perspective, conciseness, and effectiveness of the content. Your suggestions should result in a polished and well-crafted piece that effectively communicates its intended message.”
- Assistant
Here’s the refined version of your content with improvements in structure, clarity, active voice, and conciseness, while ensuring the overall message remains clear and professional:
Refined Version:
"As identified, the LogMeIn beta currently functions only on Mac devices equipped with Intel processors. It does not support devices powered by Apple Silicon chips, including the M1 and M2 series."
Improvements Made:
-
Conciseness:
- Removed redundant phrasing like "as checked" and rephrased "it is not working" to "does not support," which is more precise.
-
Active Voice:
- Replaced the passive tone ("is only working") with active voice ("functions only").
-
Clarity and Word Choice:
- "Devices with Apple silicon chips" was streamlined to "devices powered by Apple Silicon chips."
- Mentioned "processors" instead of "chips" for Intel, as "Intel processors" is a more common and precise term.
-
Professional Tone:
- Improved readability and phrasing to convey the message more effectively and formally.
-
Audience Perspective:
- Focused on delivering a clear and precise message for the audience to easily understand the current limitation of the beta software.
This polished version ensures high-quality standards by combining effectiveness, clarity, and conciseness. Let me know if you'd like further revisions or need additional assistance!
-
- User
https://www.youtube.com/watch?v=IAMyoAHYfcs
need complete transcript please
- Assistant
I currently cannot access the YouTube video you referenced (
https://www.youtube.com/watch?v=IAMyoAHYfcs) or generate a direct transcript from it, as I lack the capability to browse the internet or watch videos. However, you can easily obtain a transcript of a YouTube video by following these steps:
Steps to Get a Transcript from a YouTube Video
-
Open the Video:
- Visit the video page using the YouTube link.
-
Access the Transcript:
- Below the video, click on the three-dot menu (just below the video or near the description section).
- Select Show Transcript from the dropdown menu.
-
View the Transcript:
- The transcript will open on the right-hand side of the video.
- It will be displayed with timestamps for each section of text.
-
Copy the Transcript:
- Highlight the entire transcript (Ctrl+C on Windows or Command+C on Mac) and paste it (Ctrl+V/Command+V) into a document for further editing.
What if the Transcript Option Is Unavailable?
- The transcript option might not appear if the video creator has disabled it or if the video doesn't have automatically generated captions.
- In such cases, you can use external tools or software that supports transcriptions, such as:
- Otter.ai or Sonix: Automated transcription tools (limited free tiers available).
- YouTube Downloaders with Subtitle Support: Some tools can download video subtitles if available.
Alternative Method:
If you cannot access the transcript or want me to assist with analyzing a particular part or focus, feel free to provide a summary of the video, and I’ll be happy to help craft information based on that.
Let me know how else I can assist!
-
- User
00:00:00 here's everything that you need to learn so that you can get a job as an Azure AI engineer in this video I'm going to cover all of the Azure AI services that you need to know because the goal is to become an Azure AA engineer as fast as possible first of all Cloud engineering is in demand and is one of the best paying professions nowadays I'm sure it's going to still be relevant for years to come because more and more businesses are replacing their infrastructure with cloud services and if we pair that with AI I think we have
00:00:25 a winning combination so one of the best ways to stay relevant in the next years is to develop a Solutions on Azure about myself I've been using Azure for years I'm a Microsoft certified AI engineer and I'm a datab bricks partner Solutions architect champion and I've built a lot of production Azure data brick Solutions and that got me that title and when it comes to Azure alone I've done the same specifically leveraging their AI services so you can be certain that I know exactly what I'm talking about in
00:00:50 these videos and I'm trying to help anybody that is interested in pursuing these uh these passions of mine so hit the like button subscribe and share this if you want to become an azure AI engineer let's get straight to the point and let's look at all of the Azure AI services that you need to know so that you can get a job as an Azure AI engineer Azure machine learning Azure ml is one of the most straightforward Services because it helps with creating training and deploying machine learning models it supports managing training
00:01:17 data building models using various machine learning algorithms and deploying those trained models as web services and that's because you want to deploy those you want to integrate them into your applications the basic idea is that you have some data that is labeled that's your training data then you take that data and then you use it to train a machine learning model from that training you get a final model and then once you're happy with its performance then you deploy it potentially as a web service that you can then later use
00:01:42 through different applications and that's the gist of it Azure AI services this is actually a collection of apis sdks and services it includes services for vision for speech language decision and search functionalities we're going to cover all of them individually but the reason why Azure groups all of these Services Under one service is that you can actually access all of these AA capabilities through a single platform you don't necessarily need to manage multiple separate services or apis so building is simplified under one
00:02:10 umbrella and you only pay for the resources that you use you get better integration because you only use one set of keys and endpoints so it's easier to actually develop and integrate your AI Solutions computer vision or AI Vision AI Vision provides algorithms in order to process images and videos and then it returns information based on visual features it's used for task such as analyzing images in order to provide descriptions or to read printed and handwritten text or to recognize objects and faces the actual services that live
00:02:38 under this umbrella are image analysis spatial analysis OCR and facial recognition there's actually two more services that have crossover one can actually help with reading handwritten text and the other for face analysis going to cover them later but for example for OCR computer vision has the ability to extract text from images and PDF files for example you can use the read API to extract the the text and then you can use a text analytics service such as the text analytics API in order to perform sentiment analysis
00:03:07 on that specific text and for fa analysis you can consider this as the general computer vision service for example here's a face but here are also the faces attributes so do they have glasses do they have a mask on and many many more details that allow you to create more detailed use cases around them air vision is important for you to master if you actually want to become an AI engineer because it can be used for so many use cases and actually companies are looking to develop Solutions such as
00:03:31 image tagging or for Content moderation on social media platforms or for image search I mean look let me know down in the comments if you want a deep dive into computer vision but next let's actually look at custom Vision because this service actually allows you to easily train your own image recognition models so you can build your own model and then you can integrate it into an application and this is great because you can build both image classification and object detection models without actually needing any deep learning
00:03:58 experience and all you need is an image data set and then you can just train your model and you can deploy it as a rest API custom vision is extremely useful when you actually have a custom industry specific use case something like quality control in manufacturing or product recognition in e-commerce or object detection in surveillance systems for example as I mentioned the face service earlier this has some Crossover with computer vision but it's actually more specific because you can actually
00:04:22 identify human faces and their attributes in both images and in videos and when we think about attributes we can think about emotions or facial landmarks for verification you can actually get the head post for example and this is helpful for live lens verification in videos you can also get the facial hair you can get whether the person has glasses on or if the person wears a mask but keep in mind that all of these attributes are predictions so you're going to need some extra logic in order to use them in your applications
00:04:47 the way that it works is that it detects human faces in an image and then it Returns the rectangle coordinates of their locations and then it also returns a unique ID that actually represents the stored phase data plus of course the phase related attributes that mentioned so you can use all of them later on for operations like to identify a person or to verify a person's identity think about it in terms of access control or even personalized marketing or even sentimental analysis in market research
00:05:12 or a much simpler one to actually just tell people to take off their sunglasses or masks if they enter a location next you have to understand how to use document intelligence which was previously called form recognizer both names are actually straightforward for what it does and it's pretty obvious because with it you can extract structured data from forms and from multiple types of documents it can automatically extract information such as text and Fields from pretty much any type of document and that includes
00:05:36 invoices and receipts what you can do is to analyze the documents layout and identify key fields and sections then you can train custom models to actually recognize these layouts and Fields and then you can extract the text and pretty much any data that you want this is a very straightforward service that helps for example with automating data entry tasks text analytics this is a core Azure service that provides NLP for sentiment analysis key phrases extraction named entity recognition entity linking and language detection
00:06:04 and all of these would require a separate video in order to dive deep into all of them but in a nutshell text analytics can analyze the sentiment of text documents to determine whether the expressed sentiment is positive negative or neutral it also extracts key phrases and important phrases from text documents it can also identify and classify named entities in these text documents think about entities such as people organizations locations and dates for example and for language detect it can automatically detect the language of
00:06:32 the text document the number of use cases where you can use these Services is huge and I think these are the quick wins that you can get by using Azure AI in your company so text analytics is a must and you're going to use it paired with many other services that I'm talking about Azure AI search is a crucial service that you're also going to need to learn so that you can get a job as an Azure AI engineer because with AI search you can build scalable search Solutions this way the users of your application can find information quickly
00:06:59 I made a video about setting it up a while ago but I think AI search needs another one so we can go into all of it specific but for now let's just get the gist of it because you injest data from multiple sources and then you extract the content and then you transform it into searchable documents and then you push it into search indexes and once the data is indexed you can perform complex searches and then you can get the relevant information using Advanced queries I'm actually using AI search on
00:07:23 get badge.com and I'm using it in the question answering assistance at get that badge.com we offer both practice exam and AI assistant to help you prepare for cloud certification exams we currently have both Azure and data bricks practice exams and we're adding more and more every month so definitely check it out as if you're looking for a way to support this channel this is a way you support decision forest and you support Yourself by learning a new skill the last two services that I want to cover are the language and the speech
00:07:51 services and at the end I'm going to tell you exactly what you need to actually focus on in order to get a job as an Azure AI engineer space services are a part of the broad AI services and it actually gives us tools to add speech features to our applications and there are a lot of things that we can do with it but the first is obviously speech to text this turns spoken words into written words and it's great for writing down what somebody says and it's even better than this because you can actually recognize speakers and he can
00:08:16 also translate spoken words into many other languages and you can also do that in real time and all of these features are great when you want to capture conversations and understand them but what's even better is that you can also have text to speech and this changes written words into spoken words you can use it in virtual assistant you can use it in applications that use spoken audio or in your videos if you want to do a faces channel the nice thing about it is that it has different voices and
00:08:41 different languages that you can choose from it also lets you change how the voice sounds and I think the speech sounds very natural for pretty much these AI voices speaking about the language service I already covered text analytics a bit earlier but there's two more services that you can look into the first is the translator service and this one translates text and you can do that both in real time and you can do it in batch this is a great service for live conversations and if you can pair it
00:09:05 with text to speech afterwards then you get yourself a great use case with this pairing the last AA service that I want to talk about is language understanding this again deserves a full video about it and if you're interested let me know but overall it allows us to build custom machine learning models that can understand user commands and then it performs actions based on those commands now we have so many services that we can use as AI Engineers so what should you start with I think it actually depends a
00:09:29 lot on the use case that you choose currently Microsoft has an accelerator for pretty much every use case that you can think of and that's your starting point check out what they're already offering as Deployable templates and you can just start from there personally I suggest that you start with use cases that involve AI search and the reason for that is that many companies need this pretty much any company has a Q&A use case that they need to implement as soon as possible and that can get you
00:09:53 through the door with using these type of services and then you can build your expertise from there becoming an Azure AA engineer has never been easier today you have so many resources that can get you up and ready in no time and a lot of deep technical expertise has been abstracted the way behind these clickable interfaces whether that's good or bad that's up to you to decide but this is what we have now so what services do you want me to cover in depth let me know down in the comments and also like And subscribe if you want
00:10:19 to become an Azure AA engineer I'll see you in the next one
generate report with all key points
- Assistant
Key Points from the Video: A Comprehensive Guide to Becoming an Azure AI Engineer
This report summarizes the essential content from the video and outlines the steps, Azure AI services, and tools needed to become an Azure AI engineer.
Overview
- Objective: Learn the key steps and Azure AI services necessary to quickly become an Azure AI Engineer.
- Why This Is Important:
- Cloud engineering, paired with AI, is in high demand and offers high-paying career opportunities.
- The continued adoption of cloud services ensures that expertise in this area will remain relevant in the coming years.
Speaker’s Background:
The speaker has:
- Years of experience using Azure services.
- A Microsoft Certified AI Engineer title.
- Held the role of a Databricks Partner Solutions Architect Champion.
- Built Azure production solutions, leveraging its AI tools extensively.
Azure AI Services to Learn
To become an Azure AI engineer, the following services are essential:
1. Azure Machine Learning (Azure ML)
- Purpose: Creates, trains, and deploys machine learning models.
- Core Usage:
- Manage training data.
- Build models using various ML algorithms.
- Deploy trained models as web services.
- Process:
- Collect "labeled training data."
- Train a machine learning model.
- Deploy the trained model into applications (e.g., as a web service).
2. Azure AI Services (Suite)
- What It Offers:
- A collection of APIs, SDKs, and services organized under one platform.
- Includes functionalities for vision, speech, language, decision-making, and search.
- Advantages:
- Simplifies integration (uses the same set of APIs/keys).
- Pay-as-you-go model (pay only for resources you use).
- Centralized infrastructure for multiple AI services.
3. Computer Vision (Vision AI)
- Purpose: Processes images and videos for extracting visual insights.
- Capabilities:
- Image analysis (object/facial recognition, descriptions, etc.).
- Spatial analysis.
- OCR (Optical Character Recognition) - extracting text from images or PDFs.
- Face detection and analysis (attributes including masks, glasses, emotion, etc.).
- Use Cases:
- Image tagging.
- Social media content moderation.
- Content-based image searches.
4. Custom Vision
- Purpose: Train custom image recognition models for industry-specific use cases.
- Capabilities:
- Build models for image classification and object detection.
- No need for deep learning experience.
- Use Cases:
- Quality control in manufacturing.
- Product recognition in e-commerce.
- Object detection in surveillance systems.
5. Face API/Service
- Purpose: Detect human faces and extract facial attributes from images or videos.
- Capabilities:
- Attributes include emotions, landmarks, head pose, facial hair, masks, and glasses.
- Applications:
- Personal identification (e.g., access control, verification).
- Sentiment analysis for marketing.
- Personalization in applications (e.g., removing masks during entry).
6. Document Intelligence (Formerly Form Recognizer)
- Purpose: Automates the extraction of structured data from forms and documents.
- Capabilities:
- Extracts text and fields from receipts, invoices, and various documents.
- Custom model training for specific layouts and fields.
- Applications:
- Automating data entry tasks.
- Processing financial or legal documents.
7. Text Analytics
- Purpose: Provides natural language processing (NLP) capabilities.
- Key Features:
- Sentiment analysis (positive, negative, neutral).
- Key phrase extraction.
- Named entity recognition (e.g., organizations, locations, or people).
- Language detection.
- Applications:
- Sentiment analysis for market research.
- Text extraction and context analysis in content management systems.
8. Azure AI Search
- Purpose: Enables scalable and intelligent search solutions.
- Key Features:
- Indexes and processes searchable content from multiple data sources.
- Provides advanced search queries to retrieve relevant results.
- Applications:
- FAQ systems (e.g., question-answering assistants).
- Knowledge discovery platforms.
9. Speech Services
- Purpose: Adds speech-related functionalities to applications.
- Core Components:
- Speech-to-Text: Converts spoken words into text (supports real-time recognition and translation).
- Text-to-Speech: Converts text into spoken audio (supports multiple voices and languages).
- Applications:
- Virtual assistants.
- Real-time language translation.
- Voiceover for video content.
10. Language Services
- Purpose: Facilitates language understanding and translation.
- Key Features:
- Translator Service: Translate text in real-time or in batches into multiple languages.
- Language Understanding: Builds custom ML models that can parse and act on user commands.
- Applications:
- Custom NLP use cases for chatbots.
- Voice assistants paired with speech services.
Recommendations for Getting Started
-
Start Small:
- Focus on AI Search as a starting point. Most companies have FAQ or Q&A needs that are easy to implement but high in demand.
- Practice ingestion, indexing, and retrieving data.
-
Choose Use Cases:
- Identify use cases that align with company needs (e.g., image tagging, document structure extraction, or AI-powered search).
-
Leverage Microsoft Accelerators:
- Use Microsoft's accelerator templates as quick-start options for implementation. These offer pre-built solutions for many AI scenarios.
-
Build Knowledge Gradually:
- Familiarize yourself with easy-to-implement services first (e.g., Computer Vision, Text Analytics) before moving to more complex or custom services.
-
Hands-On Learning:
- Create a small AI project or prototype using Azure AI free-tier services to get practical experience.
Speaker’s Advice
- Becoming an Azure AI Engineer has become easier because:
- Microsoft and other providers have simplified complex AI processes through user-friendly interfaces.
- Resources and community support are widely available.
- Interactive Tools Available: Tools like exam preparation platforms (e.g., practice exams at getbadge.com) are great for building skills and preparing for certifications in Azure AI.
Conclusion & Next Steps
By mastering Azure AI services such as Azure Machine Learning, Computer Vision, Text Analytics, AI Search, and Speech Services, you are well on your way to starting a career as an Azure AI Engineer. With Microsoft's streamlined tools, numerous accelerators, and high demand for cloud-based AI solutions, now is the perfect time to pursue this career path.
Would you like deep dives into any specific Azure service or help with certification preparation? Let me know! 😊
- User
00:00:00 they're some of the world's most indon tech jobs these are synthetic idas as good as a genuine ID didn't understand the you know the the phrase it took it literally oh it even read my handwriting is this what people mean when they say that AI is going to take our jobs this CNBC video is going to tell us exactly why we need to be careful because a is just going to take all of our jobs and the only solution is to get in line and become an AI engineer because this is really in demand right now what does a
00:00:31 day in one of these roles actually look like fighting scams and fraud developing large language models from scratch this is exactly what you're going to do developing large language models from scratch to designing chat Bots they're all working with one technology AI yes that's exactly what you're going to do you're only going to work with one technology because AI is just like one thing you see this is the problem with many people they they convolute so many subtopics and subdomains and sub Technologies into one large umbrella
00:01:04 like AI you know AI is just one technology that you need to learn this is Most Wanted I'm going to show you some cool stuff Nessa yeah absolutely cool that's a that's a mask of your face yeah it's an exact replica of how how I look like Raj Mah so what does this have to do with AI I mean I understand that it it it mixes different areas and different specializations to build products but what does this have to do with what an AI engineer does wari is part of the Cyber and intelligence Solutions team at mcot in
00:01:46 his role his team develops tools to manage customer risk and prevent scams part of his role involves getting in the mind of potential frauders by creating fake identities so these are synthetic idas okay so you're creating some masks and some fake identities to do what just to put yourself into the mind of a of a criminal what what is this you're you're you're just creating personas what does this have to do with AI engineering maybe maybe they're going to tell us right now okay let's uh as a genuine ID
00:02:25 you can have these identities you can have the face mask you can have the fingerprint and then uh you are essentially replicating someone else that person can do anything with these this was done by a Japanese artist who did the mask for James Cameron movie Avatar sometimes we have to think like Bad actors to uh come up with the solution which can stop these things you have seen that the world has evolved from these mask and now the defects are coming in defects coming in okay so what does what what do defects have to do
00:02:56 with the masks that they created I mean it's just an exercise to look like they're doing something I don't I I don't really get it cuz if you want to uh prevent fraud and deal with this type of fraud detection yeah you you can create various models for fraud detection yeah but you're going to use normal machine learning and deep learning to to achieve that actually not even deep learning you're just going to use normal machine learning and specific algorithms to to see the transactions that just stand out uh you're going to
00:03:27 just create various features and then based on those features you're going to create a model and train a model to detect fraud but I don't understand what creating these masks has to do with with with anything specifically when it comes to AI engineering was not to break the technology but intent was to help uh the solution providers to enhance the level so that they can stand against these kind of attacks as well okay so we're going to create these little masks and then you're going to pretend like you're
00:03:55 the criminal and then you're going to get into the mind of a criminal and this is going to help you prevent those attacks what is this wow what did we have here yes please please please have a seat all right some people working Services some people working from a meeting room I mean who works from a meeting room did you ever really work hard and were you able to concentrate when you were in the meeting room with somebody else Joe Garcia and his team have made a game out of simulating real world security
00:04:25 conditions as well which okay we're just going to create this now another game okay so this is exactly what AI Engineers do right they create these little masks and then they create little games they hope will help clients improve their incident response processes what we have here is a project called chaos Kitty why Kitty well I have cats at home who always destroys my furniture oh yes agents of chaos for sure what exactly is chaos engineering we're going to intentionally inject some failure so that we can learn you can see
00:04:57 that there's many colorful lights and on bricks that we use to kind of represent what we have in our AWS Cloud when it's red something is wrong with the security configuration and when it's GRE it's all good on compliant oh this is great like you really need to build all of these Lego bricks and whatnot and to to replicate a real world uh AWS infrastructure so that you can uh spot the the problems in your systems I mean why why would you do that you already have that in uh in in monitoring tools and you already know
00:05:29 what systems are f I think this is just kind of like playing around you know like for example ux designers when they had these meetings in which you needed to um ideate and you need to get stakeholders involved because they couldn't really understand the technology so maybe maybe that's what they're doing right they're just creating these things so that people can understand what AWS infrastructure is but this isn't the job of an a engineer we added on a gen assistant here typical you know company security
00:06:00 policy typically without a gen assistant you know they have to look through study this policy document very deeply right we have fed this document in uh along with the Gen assistant so it that's very very nice so what they've done they used uh for example uh Azure has document intelligence so AWS has a similar system but what is this they jumped from monitoring infrastructure with some Lego blocks and some um some failure mechanisms to uh to document intelligence and processing this with Gen because now we have a gen system
00:06:35 that we don't really have to read these documents they're just jumping in from one thing to another it doesn't really make sense like if I really if I didn't know what uh n engineer uh does I would look at this video I would be so confused I wouldn't really understand like what what what are these guys doing are they like prompt engineering are they like building some Lego type of stuff are they like creating these nice masks and putting them on and pretending that they're criminals what are they
00:07:06 doing almost all this information I could go in there and ask and you can ask questions of the doc best practices going I give you some best practices fix chaos Kitty okay remediating chaos Kitty environment uh so what I'm uh what I'm understanding from this is that they have some documentation on chaos kitty yeah on some infrastructure documents and uh the Gen assistant proposes ways to to solve a failure in that system based on the documentation that's what I'm that that's what I'm assuming that it happens
00:07:45 now cuz if I'm trying to piece this together maybe maybe that's it what do you guys think you'd be able to leverage Ai and J to actually help fix the challenges uh the challenges and then they'd be they'd be more focused on the other areas that could be improved to be able to converse naturally with an AI chatbot a large language model or llm is needed l oh so you cannot this doesn't make sense this this phrase doesn't make sense so if you want to chat with a chatbot you need an llm the chatbot relies on an llm that is the
00:08:25 chatbot it's a large language model it's you don't have a chatbot without a large Lang language model or um a natural language processing solution are AI models pre-trained on vast demands of data which can understand and generate human language responses popular models include chat gbd4 and Gemini the generative AI Market is expected to grow over a trillion dollars in the next decade oh my God the trillion dollars so $1.3 trillion how are you going to get that money money get that money become an AI engineer and
00:09:02 get that money how are you going to get that money you need to create an AI solution right how are you going to create an AI solution you're going to create an AI Solution on top of these large language models the team in Singapore is developing a large language model that's catering to Southeast Asian languages Leong way speaks 14 languages fluently and he's an Ari engineer and linguist involved in developing the sea lion model we're going to look at and that's very nice so it's a nice uh example informal Indonesian we're
00:09:34 essentially asking the model our friend is sort of because of his work he he's sort of panicking you know and he's working with heart every day how can we help him best manage his work yeah but in the in that context they're saying they're using an idiom right and they're saying that um his beard is always on fire mhm exactly so so so that idium um might not be so um understandable to certain models right so let's see how they they de with understand so demo and each ofes langage models with specific understanding in a
00:10:09 specific area like for example I think this is southeast Asia so they're they're customizing they're building probably on top of an already existing large language model probably something open source and then what they're doing they're just adding they're fine-tuning that model with with specific language and understanding topics from that region and the fact that this guy speaks uh 14 languages that's really really irrelevant language model so on the left we have C lion our model and we have three other models on the right um we
00:10:41 can see that it it didn't understand the you know the the phrase it took it literally exactly like it's thinking about setting someone on fire oh no it's also in the same tone exactly it's casual informal it's still colloquial whereas this one is still sticking to like they're comparing various output kind of exactly and and and that's really what we want to achieve as well now we are seeing that some of these models out there they are not able to sort of handle Multicultural context um and and that's sort of understandable
00:11:10 because um they're building those models all the like for example jpt right it understands multiple languages it understands Indonesian and all that all those languages so um I'm guessing like they just want to fine-tune specific models to just be um better on specific benchmarks in a particular area so they're trying to uh to create specialist models for a specific region so yeah I I think that that's quite interesting particular audience right for us in Southeast Asia we need to operate within this region handling
00:11:44 our languages and cultures so this is why we decided to build sea lion yeah Vincent o works as a senior specialist Solutions architect at Amazon web service okay so now they're jumping to another thing and his projects involve lers do because I'm thinking like eventually we're going to find out what an AI engineer does no let's let's see if uh if we find out what an AI engineer does I based on human prompts to create personalized experiences St gen was a project that we did with the national libr board of attire Singapore but we
00:12:17 wanted to reinvent the future of the libraries gen Ai and AWS technology to really experience where by young children adults they can put in a series of inputs uh the selection and a brand new book will be created on a spot for them I see so this is just an application of AI like just generating images and potentially even short clips with um by by prompting based on based on a book so this is this is just an application of creating children's books or whatever so it's not really like okay how are you
00:12:51 how are you doing this what you call the prompts promps right that you're actually sending back to the uh large language model this is just the very beginning of the art of the possible and it's going to be amazing how people will leverage gen to unleash their the the reality about many AI engineer jobs you're just literally creating apps that are built on top of large language models okay so it's it's kind of like a wrapper around I don't know open AI or clae or again like stable diffusion if um if we're talking
00:13:26 about those uh children's books or whatnot it's kind of that plus a hopefully a fancy looking uh UI but you see they're kind of referring to AI as literally just prompt engineering and making some API calls and trying to put together a nice little app it they they don't really go into what what skills an AI engineer should have in order to achieve a specific goal is this the the job that is going to get paid I don't know like $400,000 you know for engineer to do this to what write API calls is this the future is this the
00:14:05 future of AI engineering I don't know extended level of creativity all right so you just need to be creative and just try to find ways to to create better prompts yeah that's all an AI engineer does if this is the case guys like this is the case this job is going to just crash when it comes to to um the amount of money that you can make by being an a engineer like it's it's insane they're simplifying AI engineering and pretty much machine learning and a deep learning to to just making API calls and this is the uh
00:14:49 the job that is supposed to be in demand part of those new jobs that are created as part of AI is uh is a role or a skill set which is actually prompt engineering which didn't exist before okay and um when you look at prompt engineering you don't NE you don't they don't need to be technical see understand 1.3 trillion how put those prompts you're making 1.3 trillion yeah this is projected to grow in the next years and you don't need to be technical get in line guys and get that money get that money because you don't
00:15:26 need to be technical all you need to do is just type type and uh create some nice prompts there you go $400,000 you can just make that money like that yeah oh my God they're simplify simplifying it so much just awful uh to maximize and Leverage The Power of a l language mod there is a growing demand for AI Specialists and here at AI Singapore how can you be an AI specialist if all you do is just prompt engineer or if you just fine tune with certain um certain uh value Pairs and trying to um to to
00:16:03 improve uh a large language mges marginally on a specific I don't know culture or language and whatnot like it's not that hard no you're not really building a large language model from scratch and this is exactly what these guys have been saying up until now right they said it at the beginning of the video but I was looking at uh at the sea lion right and the key features of sea lion right the training strategy it's built on top so the version two was built on top of Lama 3 and uh the V3 was built on top of Jimma 2 okay
00:16:40 so um initially I think they were pre-training it from scratch they say that the the first version uh that that was initially trained but then they just train it on top of a larger language model okay so probably in 2017 when they started they they started it with this big goal but then when they realize like oh we can just train it on top of an already existing large language model we're just going to use that let's just drop what we had and let's just use this so it's um it's interesting but anyway the rest of weight's Team hailing
00:17:17 from all over southeast Asia including Vietnam the Philippines and Thailand are working on making sea lion more inclusive you are the one who brought everyone into this team more or less more or less my dream is eventually we have everybody from all the countries in isn't this like some kind of a promo or like a sponsored video cuz why are they just presented here AI singap apprenti program one application or just one or two applications and then they're just pushing them like that and making a a
00:17:47 very general video topic on it like this is what AI Engineers do what do they do I didn't find out what AI Engineers do I mean what's the skill set do you need to know math do you need know statistics uh what type of uh how does the day in the life of an AI engineer look like maybe it looks like just being in a meeting room and just uh putting some Lego blocks together creating some nice little masks maybe that's what the AI Engineers do right and they you just prompt away um and um maybe that's it
00:18:21 roll the pool locally even those considering midc career switches in fact some of the sea lion team graduated as AI appren IES after switching disciplines such as Tai who studied finance and wayt who was previously a pharmacist so you see the whole point is to anybody can become an a engineer right from any from any domain yeah just just become an AI engineer it's super easy cuz look if this if it's so simplistic that you can just do some prompt engineering and just create some uh API calls to a large language model
00:18:56 you know like a foundational model that any body can do it if anybody can do it then what's the future of the job you tell me we're not really alone in trying to tackle low resource languages optim all the stuff that we do is fully open source as well and it's really shed with the public in general so that everyone can benefit everyone is working on different aspects of the large language model we get to learn um about it with each other it does feel a bit like mini asan something minian yeah there so much
00:19:29 to do in this F yeah and AI you know is one of those Industries where you can make a mider switch and that's the thing you need now you need to make a midcareer switch Yeah uh they're trying to get more people into AI right because this is the new shiny object AI cuz nobody else is hiring now so everybody wants AI absolutely and I'm a good example Nessa ret started his career in the mobile and semiconductor industry and made the switch over to AI into 2014 that's good I have seen that Journey but
00:20:01 in 2014 he didn't make the switch to AI he made the switch maybe to some machine learning yeah because the whole AI Trend started uh to grow in the last couple of years no now everything is AI every company is implementing some AI some generative AI into their products past 20 years or so where AI has really and it's becoming an inherent part of our lives when you ask people to use digital ecosystem there are side effects according to the Global anti-cam Alliance more than 1 trillion dollars are lost to scams every year affecting
00:20:40 two billion victims R's work at mastica includes working with banks and governments to apply intelligent AI systems to predict this has been happening for a long time use machine learning to detect fraud is very straightforward right you have various uh algorithms like for example tree based algorithms that can detect fraud yeah we just need to create uh features and try to spot uh outliers yeah cuz in the end that's that's what you're doing is you're doing anomaly detection in order to uh spot potential fraud you're
00:21:15 trying to detect anomalies um but yeah now it just got this just got lumped together with AI so is this AI is prompt engineering AI what is it where the scams are taking place let's take the example that you giving me 100 in a in a day something is suspicious sound alarm Bells exactly we train the models to detect the behavior to detect the patterns and then do the risk assessment and save uh you being sending money to
00:21:51 me welcome to the Master Card experience whenever a person Taps the card our AI models kicks in then we give a risk assessment to the issuing institution that what do we feel about this transaction you can see the screen here showing that transactions are getting deined where do you see the state of AI in the next couple of years station with the llms so that's llm is essentially the large language models okay large language models form a big part of the user experien solutions that Joe works on for AWS take reading
00:22:26 inv you want to like introduce something else in your business a new form a new process you have to reevaluate the technology that in with Gen Solutions yeah you know because the large language model it can actually parse these okay so you're just using computer vision to uh and OCR actually this is OCR right you're using OCR to read uh text parsing refers the process of breaking down a user's input into smaller pieces and analyzing each piece to determine its meaning I'd like to know um if there is
00:23:01 a proper signatory mhm oh it even read my handwriting so it knows that it's my terrible handwriting that it's Michael Garcia yeah cuz you lab did you see yourself ending up where you are today even at a young age I knew that in some form way or um that I was going to be involved with technology what did you start out studying I graduated computer science I went through a variety of jobs I even founded a company before startups were fashionable to day with AI when you're thinking about redefining
00:23:32 experiences you're not going to get to it on the first yeah you're never going to get to it on the first try right you need to create multiple Solutions and then eventually maybe you're going to create um a solution that actually solves a problem right you're always trying to do that and and uh this is something quite good you know in the end you're not just studying computer science just to become an employee and just get a job as a specific um I don't know machine learning engineer engineer uh programmer
00:24:03 software engineer whatever no the you need to have a higher goal than that and listen even AI engineering right in the end this is like super long ambition right you know is not uh tan language the foundations of this right the foundations of this are um the foundation uh foundations of machine learning of maths statistics right you need to understand how to code right you can choose that know python you can choose um a language that that that can get you to where you want to be um but in the end you see you need to have a
00:24:38 higher goal right you're not just going to be a prompt engineer okay you need to think higher than that like hundreds of dialects Bal javanes you know visayan so so we hope to get those represented as well eventually how soon can we get there our hope is that it will trigger similar mov movements across the region outside of South this is yeah very very good that they're trying to to push these things yeah very strong right we have similar situation in India we have similar situation in Africa where there's a bit
00:25:12 of under representation if I'm able to save some it's normal CU in the end like all these large language models are trained on um freely available data from the internet right and the majority of um of Articles of videos of everything right they're in they're in English so therefore English language is a lot better represented than many of these narrow dialects and uh and smaller languages it's just normal yeah but with time uh I'm sure I'm sure this will change and and um and and all of these initiatives will be more successful
00:25:47 lifelong savings I mean it's itself uh that makes my day is this what people mean when they say that AI is going to take out jobs let's see it's going to make something more efficient so that people can actually concentrate on what matters more either for the business or for customers what matters more cuz listen we're going to do this hard this these things for you yeah and now you can focus on other things what things this is kind of the promise of AI right like a yeah we're just going to automate
00:26:17 all of these things and now you can just focus on something else the the the problem that I see with um with AI and of course we can talk about this in in in in another video but the the problem that is see now with AI is that it's going to make people rely so much on it that people won't be able to think without relying on on a large language model so if you if you don't do that right you're not going to be able to think about other things as well right you're just going to become dependent on these large Corpus of
00:26:58 historical information that is AI so the creativity that everybody's talking about the the the the solutions that there everybody's talking about yeah you're going to be able to create new Solutions and great stuff in the future but if your brain is going to be so slow that it cannot even think without AI slowly that creativity is going to fade away because you're only going to rely on things that were already built and as you can see even nowadays I don't see much inovation happening in AI I only see Innovation from a couple of big
00:27:33 companies but the majority of small uh players and individuals like me and you right they're um they're just leveraging these these tools and personally I think eventually we're just going to have two or three big companies that are going to provide AI Solutions and everybody else is just going to use those systems so I don't really I don't really see um creativity uh growing and getting um more widespread I actually think a lot of people will be a lot more um lazy I think they're going to be a lot more
00:28:08 lazy but what do you think what does an AI engineer do
- Assistant
Key Insights and Report: What Does an AI Engineer Do? (Based on the Video Transcript)
This report summarizes the key observations, tasks, misconceptions, and realities of AI engineers based on the video transcript. It not only describes specific points raised in the video but also provides clarity on what AI engineers do in practice.
Introduction: What’s the Buzz Around AI Engineering?
Artificial Intelligence (AI) is a rapidly growing field with applications spanning industries like fraud detection, language processing, personalized user experiences, and more. AI engineering is presented as a promising field of work, but the video transcript raises critical questions about the realities of this profession.
While AI engineering holds potential, the transcript highlights major perceived misconceptions, overlapping domains, and a lack of understanding regarding the tangible skills required for AI engineering.
Key Themes Covered in the Video
1. Misconceptions About AI Engineering
- The video humorously points out the oversimplification of AI engineering:
- Some portray AI engineering as "just creating apps with prompts" or "making API calls to large language models (LLMs)".
- Task descriptions such as building chatbots, creating fake masks, interpreting handwriting, or arranging LEGO blocks to simulate AWS infrastructure lack clarity about the actual engineering and technical skills necessary in AI professions.
- “Generative AI” (e.g., large language models like ChatGPT) is often incorrectly equated to all of AI, which overshadows other advanced areas of AI engineering like deep learning, complex model development, and infrastructure design.
2. Real-Life Tasks Performed in AI Engineering
-
Fraud Detection Systems:
- AI engineers develop systems that analyze transactional anomalies using machine learning models.
- Fraud detection primarily involves:
- Anomaly detection using machine learning algorithms.
- Feature engineering to identify unusual transactional patterns.
- Tools like tree-based algorithms are deployed to spot outliers in user behavior.
-
Document Analysis:
- Tools like Azure's Document Intelligence and similar systems convert unwieldy, unstructured text (from receipts, invoices, etc.) into structured formats using OCR (Optical Character Recognition).
- Engineers build pipelines around structured data, paired with APIs like Azure’s language services, to process documents more efficiently.
-
Development of Specialized Language Models:
- Teams fine-tune pre-trained language models (e.g., LLaMA, GPT) to cater to local contexts, languages, or cultural idioms.
- Example: “Sea Lion” (a fine-tuned large language model for Southeast Asian dialects) builds upon existing model architectures instead of training from scratch.
- Use Case: Idiomatic or cultural phrases (e.g., "his beard is on fire") are tailored for specific regions and implemented into conversational AI systems.
- Teams fine-tune pre-trained language models (e.g., LLaMA, GPT) to cater to local contexts, languages, or cultural idioms.
-
Security Tools and Chaos Engineering:
- Engineers simulate failures (e.g., "Chaos Kitty" project) in cloud infrastructure (like AWS) to test incident response, security measures, and system resilience.
- The chaos engineering approach involves injecting controlled failures into live systems to improve security configurations.
-
Generative AI Applications:
- AI engineering work often involves building user interfaces (UI) in applications that use LLMs for tasks like:
- Generating custom children’s books.
- Creating personalized experiences by combining LLM outputs with user prompts.
- These systems sit atop established infrastructure, built with APIs like those offered by OpenAI or AWS.
- AI engineering work often involves building user interfaces (UI) in applications that use LLMs for tasks like:
3. Key Skills Required for an AI Engineer
Contrary to the oversimplified notion that AI engineering revolves around prompting or app development, the core skills include:
-
Programming Acumen:
- Essential knowledge of languages like Python, R, or other ML-friendly tools.
-
Mathematics and Statistics:
- Strong foundational knowledge in probability, linear algebra, matrix manipulation, and statistical techniques for building and tweaking machine learning models.
-
Machine Learning (ML) and Deep Learning (DL):
- Understanding both:
- Traditional machine learning techniques (e.g., regression, decision trees).
- Deep learning concepts for model training (e.g., neural networks, convolutions).
- Understanding both:
-
Data Processing and Feature Engineering:
- Skills in cleaning, transforming, and preparing datasets.
- Creating custom features for analysis and developing specific data pipelines.
-
Cloud Platforms & APIs:
- Familiarity with AI technologies and infrastructure tools such as:
- Azure AI Services (e.g., Document Intelligence, Vision, LLM integrations).
- AWS AI/ML Services.
- TensorFlow, PyTorch for custom solutions.
- Familiarity with AI technologies and infrastructure tools such as:
-
AI Model Deployment:
- Deploying and scaling models for real-world applications (e.g., fraud detection, personalized recommender systems).
**4. The Rise of Prompt Engineering
- While the video emphasizes prompt manipulation for LLMs (e.g., ChatGPT), this is just one small area under AI engineering:
- Engineers often couple LLMs with practical applications by integrating them into apps, backends, and customer-facing dashboards.
- Prompt engineering, while useful, requires no deep programming knowledge and risks oversaturation in the labor market.
5. The Future of AI Engineering
- The video transcript acknowledges the hype surrounding AI engineering. It hints at potential challenges:
- As tools like OpenAI APIs simplify the development process, AI engineering runs the risk of being commoditized.
- With pre-packaged AI solutions, the need for foundational expertise in ML and statistics could drop as businesses mostly leverage third-party solutions.
- Despite these risks, AI remains valuable for mid-career switches (e.g., pharmacy, finance professionals joining AI).
Insights and Conclusions
What an AI Engineer Actually Does
AI engineers work on a variety of complex and technical tasks. Examples include:
- Building algorithms and deploying machine learning models.
- Developing tailored AI applications for industries (e.g., healthcare, finance, retail).
- Performing research and customization of existing AI tools (like fine-tuning LLMs such as GPT or LLaMA).
- Ensuring scalability and reliability with cloud tools like AWS or Azure AI Services.
- Solving real-world problems (e.g., fraud detection, sentiment analysis, document processing).
Common Misconceptions
- AI is a single tool: Many wrongly believe it’s "one technology" — AI spans various domains like NLP (Natural Language Processing), CV (Computer Vision), predictive analytics, and robotics.
- AI engineering is prompt writing: Typing clever prompts into LLM chatbots is only a tiny part of AI engineering. It lacks substance in the long run.
- Generative AI dominates all AI technology: Businesses must still rely heavily on traditional machine learning for applications like fraud detection.
Warnings from the Speaker
- Heavy reliance on LLM-based tools like GPT poses a danger. Overdependence on automated models may hinder human creativity and technical innovation.
- The reduced demand for deep, complex programming work could mean that only a few large companies (OpenAI, Google, Microsoft) will build and control AI systems, while smaller players simply use their APIs.
Recommendations for Aspiring AI Engineers
-
Learn the Basics:
- Mathematics (linear algebra, calculus, probability).
- Programming (Python, TensorFlow, PyTorch).
-
Focus on Scalability:
- Learn about deploying solutions via cloud platforms (Azure, AWS, GCP).
-
Don’t Rely Solely on Generative AI Hype:
- Explore legacy AI systems like fraud detection or recommendation models.
-
Build a Portfolio:
- Demonstrate knowledge by building independent projects or contributing to open source.
-
Mid-Career Switch Is Possible:
- AI engineers can come from non-CS fields (e.g., finance, healthcare) by focusing on real-world applications of AI.
Closing Thoughts
AI engineering is a technical, innovative, and rewarding career, but it requires substantial expertise in math, coding, and problem-solving. Simplifying the profession to mere "prompt engineering" or "creating small apps" undermines its complexity and skill requirement.
The real challenge is balancing innovative applications of AI with deep technical knowledge, while avoiding dependency on out-of-the-box tools that may sap creativity. What specific projects or skills will you work on to stand out as an AI engineer?
- The video humorously points out the oversimplification of AI engineering:
- User
- Assistant
I'm sorry, but I cannot access specific YouTube URLs or videos directly as I do not have browsing capabilities. However, I can help you with strategies for obtaining transcripts or summarizing YouTube video content:
How to Get a Transcript from a YouTube Video
-
Using YouTube’s Built-In Transcript Tool:
- Go to the YouTube video.
- Click the three-dot menu (below the video player).
- Select "Show Transcript".
- You can copy the transcript and analyze it as needed.
-
Third-Party Tools for Transcription:
- Use transcription tools like Otter.ai, Rev.com, or Sonix to generate transcripts by uploading the audio/video link.
-
Summarize the Content:
- If you're looking for a summary of the video, you might explain its key points or main idea, and I can help you rewrite or analyze them.
If you'd like, describe the video, share any key details, or paste the transcript here, and I’ll help you create a detailed summary or analyze the material for important insights! Let me know how I can help! 🙂
-