DIFINES AI
RAG-based AI consultant that answers user questions about the DIFINES ecosystem using markdown knowledge resources, Supabase pgvector search, and Groq-powered LLM responses.
Role
Full Stack AI Engineer
Technical Highlights
- Built a RAG chatbot for the DIFINES consultant page using markdown documents as the knowledge source
- Implemented Supabase pgvector storage and similarity search for retrieving relevant ecosystem documentation
- Integrated Groq Llama 3.3 70B for fast AI responses and Gemini embedding-001 for document embeddings
- Created an ingestion workflow to chunk markdown files, generate embeddings, and store searchable knowledge in PostgreSQL
- Designed the assistant UI to match the existing DIFINES landing page style and provide a smooth chat experience
Impact
Improved user understanding of the DIFINES ecosystem by enabling instant AI-powered answers based on verified project documentation.
- React
- TypeScript
- Supabase
- PostgreSQL
- pgvector
- AI SDK
- Groq
- Llama 3.3 70B
- Gemini embedding-001
- RAG
- Markdown Knowledge Base
- Tailwind