all projects
PhysAI
RAG-powered robotics education platform with AI chatbot. Combines structured content on robotics and physical AI with intelligent content querying that provides referenced answers.
The problem
Robotics and physical AI content is fragmented across papers, docs, and videos. Learners waste hours hunting for the right source instead of actually learning.
What I built
A structured knowledge base paired with a RAG chatbot. Users can browse curated content on robotics topics, and at any point ask the chatbot a question — it answers from the knowledge base and cites the exact sources.
Stack & decisions
- Agents: OpenAI Agents SDK for tool-calling and retrieval orchestration.
- Backend: FastAPI — fast iteration, clean async story.
- Frontend: Next.js with streaming responses so answers feel instant.
- Retrieval: vector search over structured content, with source-aware citations in every answer.
Outcome
Shipped a working prototype that demonstrates how domain-specific RAG beats generic chat for technical learning — citations and curation matter as much as the model.