Sarthak

AI Video Assistant with Retrieval-Augmented Generation (RAG)

Built an AI-powered video assistant that enables users to ask natural-language questions and receive context-aware answers from video content. • Designed the complete RAG pipeline including video processing, transcript generation, text chunking, embedding generation, semantic retrieval, and LLM-based response generation. • Integrated OpenAI, Llama, and Gemini models to generate responses grounded in retrieved transcript information. • Improved retrieval relevance through chunking strategies, embedding models, and prompt engineering.

Multi-Agent AI Research System using LangChain

Built a multi-agent research system that automates information gathering, analysis, summarization, and report generation. • Designed specialized Research, Planner, Summarizer, and Reviewer agents to divide complex research tasks into structured workflows. • Implemented agent orchestration, communication, and custom tool routing using LangChain. • Integrated LLMs including OpenAI, Gemini, and Llama for intelligent task execution and content generation. • Connected web search APIs, document loaders, and vector databases to enable external knowledge retrieval and research workflows.

Gender and Age Prediction using machine learning

• Developed a machine learning system for predicting age and gender using behavioral and facial-expression-related inputs. • Applied machine learning techniques to build predictive models and evaluate their performance. • Explored potential applications of predictive modeling in user-experience and analytical use cases.