Chat with YouTube Videos | Nikhilesh Narkhede

Project Overview

Designed and built an end-to-end conversational AI system enabling users to ask natural language questions about long-form YouTube videos
Extracted and processed video transcripts, performing text chunking and preprocessing to support efficient semantic retrieval
Generated dense vector embeddings using Sentence Transformers and HuggingFace models to capture contextual meaning
Implemented semantic search pipelines using FAISS and ChromaDB to retrieve the most relevant transcript segments in real time
Integrated Large Language Models (LLMs) via LangChain with OpenAI and Anthropic APIs to generate accurate, context-aware answers (RAG architecture)
Developed an interactive Streamlit application to deliver low-latency, user-friendly conversational experiences
Demonstrated applied NLP and GenAI system design, combining retrieval, embeddings, and generation into a production-style workflow

RAG-based conversational AI with semantic search using FAISS and ChromaDB, LLM integration via LangChain, and interactive Streamlit interface.

Python NLP Conversational AI RAG LLMs LangChain HuggingFace FAISS ChromaDB Vector Databases Streamlit API Integration