Dan Chen – AI Portfolio Chatbot

An AI-powered interactive portfolio that lets visitors chat with an intelligent assistant trained on my real resume, projects, and experience. Built using Retrieval-Augmented Generation (RAG) to deliver grounded, accurate responses.

Live Demo: danchen.dev

---

How It Works

This project implements a Retrieval-Augmented Generation (RAG) pipeline that connects structured professional data with natural conversation:

1. Embedding Creation: Resume and project content are embedded using OpenAI’s text-embedding-3-small model.
2. Vector Storage: These embeddings are stored in Pinecone, enabling fast, high-dimensional semantic search.
3. Query Processing: When a user asks a question, it’s converted into an embedding and compared against stored vectors.
4. Context Retrieval: The top matching entries are retrieved as context for response generation.
5. AI Response: OpenAI’s GPT-5-mini model generates conversational, context-grounded answers referencing the retrieved data.

This ensures highly relevant, fact-based answers grounded in my actual background.

---

Tech Stack

• Frontend: Next.js, React
• Styling: Tailwind CSS
• AI / ML: OpenAI API (GPT-5 + Embeddings)
• Database / Vector Store: Pinecone
• Deployment: Vercel
• Version Control: GitHub

Architecture Overview

User Query
   │
   ▼
[Embedding via OpenAI API]
   │
   ▼
[Vector Search in Pinecone]
   │
   ▼
[Top Matches Retrieved as Context]
   │
   ▼
[GPT-5 Response Generation]
   │
   ▼
Response → Chat UI (Next.js)

---