Yuvraj Singh

This project is an intelligent YouTube assistant that allows you to query any YouTube video and get meaningful, context-rich answers based solely on the video’s transcript. It uses LangChain, Google Gemini AI, FAISS, and Streamlit to create a vector-based retrieval-augmented generation (RAG) system.

This project is an AI-powered tutoring application designed to provide personalized assistance across various subjects. The core of the system is a sophisticated multi-agent architecture where a central Tutor Agent intelligently delegates user queries to specialized sub-agents, such as a Math Agent and a Physics Agent. These sub-agents are equipped with specific knowledge and tools to handle domain-specific questions effectively.

The application utilizes concepts from Google's Agent Development Kit (ADK) to build the agent hierarchy and interaction patterns. The natural language processing and generation capabilities are powered by the Gemini API, enabling the agents to understand and respond to user queries in a human-like manner.

• Built a CNN model for steering angle prediction using Nvidia’s research.

• Applied OpenCV-based augmentation on 500+ images; achieved 94% test accuracy.

• Integrated the model into a simulated environment for real-time decisions.

The SmartMerge Assistant is a Python application designed to facilitate intelligent branch merging on GitHub. It leverages AI to analyze changes between branches and provides a user-friendly interface for managing merge operations.

My contribution in this project was the API integration , backend development and error handeling.