The Portfolio AI Chatbot is an interactive assistant I built to give visitors a richer, more conversational way to explore my background. Instead of just reading static pages, users can ask about my skills, projects, and career experience and get instant, context-aware answers.
The bot is powered by Next.js and OpenAI’s GPT-4-mini model, wrapped in a custom retrieval-augmented generation (RAG) layer. I preprocess and chunk my résumé data, then rank the most relevant snippets with a lightweight similarity search so the model always answers with accurate details. A streaming API and edge-runtime functions keep responses snappy, with median round-trip times under 200 ms.
For deployment I built a fully self-contained widget that can be embedded via iframe on any site—my own portfolio or elsewhere—without style collisions. Strong Content-Security-Policy headers and carefully scoped sandbox permissions keep the integration secure while still allowing real-time interaction. The result is a smooth, low-latency chat experience that turns a static portfolio into a dynamic, always-on conversation.