Skip to content
TheChatArchiveThe Chat Archive
Perplexitysonar-reasoning-pro-high· PerplexityHugging Face LMArena Search Arena 24K

can you help me with the projects that are limited and I can work on in my Btech final…

From a Chatbot Arena battleThe same person sent this prompt to two anonymous models at once. Compare it with gpt-4o-search-preview-high-loc’s answer.
  1. User

    can you help me with the projects that are limited and I can work on in my Btech final year with the topics of LLM, RAG, Agentic AI, MCP etc

  2. Perplexity Assistant

    BTech Final Year Project Ideas for LLM, RAG, and Agentic AI

    As a final year BTech student looking to work on cutting-edge AI technologies, you have several exciting project options that combine Language Learning Models (LLMs), Retrieval Augmented Generation (RAG), and Agentic AI. These projects are both manageable in scope for a final year project and technically significant enough to showcase your skills.

    LLM-Based Projects

    1. Resume Analyzer

    Build an intelligent system that analyzes resumes to provide feedback and improvement suggestions. This project involves preprocessing resumes, fine-tuning LLMs like BERT on labeled resume datasets, and implementing classification techniques to categorize resumes based on job relevance, skills, and experience[10].

    Technical requirements:

    • Python programming
    • Natural Language Processing (NLP)
    • Fine-tuning of pre-trained LLMs
    • User interface development

    2. Text Summarization System

    Create a system that can automatically generate concise summaries of articles, research papers, or news content. This project demonstrates the power of LLMs to understand context and extract the most important information[9][10].

    Technical requirements:

    • Python with NLP libraries
    • Text preprocessing techniques
    • LLM implementation
    • Evaluation metrics for summarization quality

    3. Dialogue Summarization

    Develop a system that can listen to or read conversations and generate meaningful summaries. This is particularly useful for meeting notes, customer service interactions, or podcast transcripts[10].

    Technical requirements:

    • Speech-to-text (optional)
    • Dialogue processing
    • Fine-tuning techniques like LoRA
    • Prompt engineering

    RAG-Based Projects

    4. Customized Question-Answering System

    Build a Q&A system that retrieves information from specific knowledge sources to answer user queries with accurate, contextual information. This system combines the strengths of retrieval mechanisms with generative capabilities of LLMs[9].

    Technical requirements:

    • Vector databases (like FAISS)
    • Embedding techniques
    • RAG pipelines implementation
    • Sub-question query engines

    5. PDF-based Question-Answering System

    Create a system that allows users to upload PDF documents and then ask questions about the content. The system retrieves relevant sections from the document and generates natural language answers[10].

    Technical requirements:

    • PDF parsing libraries
    • Text chunking strategies
    • Vector embeddings
    • Semantic search implementation

    6. Knowledge Retrieval Engine using Langchain

    Develop an engine that can query various sources like Wikipedia and academic repositories to retrieve and synthesize information. This project demonstrates the power of combining external knowledge with LLM-generated responses[9].

    Technical requirements:

    • Langchain framework
    • API integrations
    • Vector databases
    • Prompt engineering

    Agentic AI and Combined Projects

    7. Customer Support AI Agent

    Build an intelligent agent that can handle customer support inquiries by accessing relevant knowledge bases, generating appropriate responses, and maintaining conversation context[10].

    Technical requirements:

    • Conversation state management
    • Intent recognition
    • Knowledge base integration
    • Response generation

    8. Integrating Agentic RAG with MCP Servers

    Create a system that combines RAG capabilities with Multi-Channel Provisioning (MCP) to enable an AI agent to retrieve knowledge via MCP servers and incorporate it into the generation pipeline. This advanced project demonstrates how to build architectures that allow AI agents to reason about how to retrieve better information[7].

    Technical requirements:

    • Agent (LLM) with planning logic
    • MCP server setup
    • Vector database integration
    • MCP client interface development

    9. Cybersecurity Intelligence System with RAG

    Develop a multi-agent system that gathers real-time threat intelligence, analyzes vulnerabilities, and recommends mitigation strategies using RAG techniques to provide accurate and contextual information[10].

    Technical requirements:

    • CrewAI or similar frameworks
    • Threat intelligence APIs
    • LangChain integration
    • Report generation

    10. FAQ-System using RAG

    Build a system that automatically generates and answers frequently asked questions by retrieving information from relevant knowledge bases and using an LLM to formulate natural responses[10].

    Technical requirements:

    • Document processing
    • Question generation techniques
    • RAG implementation
    • Response quality evaluation

    Implementation Considerations

    When selecting your project, consider:

    1. Data availability - Ensure you have access to necessary datasets or can create them
    2. Computing resources - Some LLM projects require significant computational power
    3. Technical scope - Choose a project that matches your programming and AI knowledge
    4. Real-world application - Projects with practical use cases will be more impressive
    5. Documentation - Plan to thoroughly document your approach, methodology, and results

    Each of these projects offers a solid foundation for showcasing your skills while exploring the exciting intersection of LLMs, RAG, and agentic AI technologies in your final year of BTech studies.

Responses were produced in different, uncontrolled contexts (system prompts, settings, prior turns). This is not a benchmark.