Live proof of skill
Ask my AI assistant.
This is a real retrieval-augmented assistant, the same pattern I build for clients. It's grounded in my own materials and will tell you when something isn't covered rather than guess.
Grounded in Josh's résumé and case studies. Answers only from his materials. It will say so when something isn't covered.
How it works
01
Retrieve
Your question is embedded and matched against my résumé and case studies in a pgvector store.
02
Ground
The most relevant passages are passed to Claude with instructions to answer only from them.
03
Cite & trace
The answer streams back with its sources, and every call is traced for observability.
Stack: Python · FastAPI · pgvector · OpenAI embeddings · Claude (Anthropic) · Langfuse. Deployed on AWS.