Built an intelligent customer support assistant that answers user queries using Retrieval-Augmented Generation (RAG). The system indexes a company FAQ PDF into a Chroma vector database using Mistral embeddings and retrieves relevant context before generating responses with an LLM. It also maintains conversational memory to provide context-aware answers across multiple interactions.

Developed a production-style multi-agent scheduling assistant using LangGraph. The application routes user requests through specialized agents that understand booking intent, validate available time slots, maintain conversation state, and simulate appointment scheduling. The workflow demonstrates stateful agent orchestration with conditional routing and persistent memory.

A data-driven football scouting platform that analyzes player statistics to identify players with similar playing styles and performance profiles. The project is designed to help scouts and analysts quickly discover potential alternatives or recruitment targets based on statistical similarity, moving beyond single stats to compare players across multiple performance attributes at once.