01
RAG · vector search
Project one — RAG doc search
Placeholder. What it does, what it replaced, the number that improved.
- FastAPI
- pgvector
- React
Open for freelance · replies in 24h
Fullstack
engineer
+ Gen AI
I build the API, the UI, and the retrieval layer that makes it smart.
Nine years across Python, Java and PHP backends and React front ends. Now wiring retrieval, agents and evals into the same systems — measured, not magic.
- Python / FastAPI
- Java / Spring
- PHP / Laravel
- React / Vue / Node
- iOS / Android
- MySQL / Postgres
What I’m shipping
02
Multi-tenant · event-driven
Project two — billing platform
Placeholder. Scale, constraints, trade-offs. Boring on purpose.
- Spring Boot
- MySQL
- Vue
03
Agents · tool use
Project three — support agent
Placeholder. Tool calling, guardrails, human in the loop.
- Laravel
- Node
- Redis
Gen AI — learning log
01
Retrieval & vector search
pgvector, hybrid search, chunking that doesn’t lie.
Shipped
02
Agents & tool use
Function calling into real APIs, with a kill switch.
Shipped
03
Prompting & evals
Building a small eval harness before trusting output.
In progress
04
Fine-tuning
LoRA on small open models. Early days, notes public.
Learning