My AI assistant — AI-Powered Portfolio Chatbot

Built and deployed a production-grade conversational AI agent as a recruiter-facing portfolio asset, designed to answer questions about my experience, projects, and technical background in real time. The system is architected around LangGraph's StateGraph primitives, enabling a multi-step agentic reasoning loop rather than a simple single-shot retrieval pipeline. Project and experience data is embedded and stored in Pinecone as a vector index, with a FastAPI backend orchestrating retrieval-augmented generation (RAG) queries against the index and routing responses through the LangGraph agent. The application is deployed on Render with a keep-alive mechanism to eliminate cold-start latency, rate limiting to prevent abuse, and a Resend-powered email capture flow to log recruiter interactions. The entire stack — from embedding pipeline to agent logic to deployment — was built independently, serving both as a live demonstration of production LLM engineering and as a practical tool that lets anyone visiting the portfolio have a natural conversation with my work history.