I build AI that works in production.
GenAI engineering end to end — agents, retrieval pipelines, and the production backends that hold them together. No prototypes that break at scale.
// What I Forge
Three pillars. Each one battle-tested across real products, real users, real traffic.
Agentic Applications
Multi-agent systems built on LangGraph. Tool-use pipelines, state machines, and workflows that stay reliable when the LLM doesn't.
Retrieval & Knowledge Systems
Search that understands context. RAG pipelines, hybrid vector + sparse retrieval, and document intelligence tuned for production latency.
Production Backends
FastAPI microservices with the observability, CI/CD, and cloud plumbing that keeps AI systems running. Multi-cloud across AWS, Azure, and GCP.
// Beyond the Core
Five years across QA, backend, ML, and freelance builds left scar tissue in places GenAI engineers rarely go. The breadth compounds.
// The Approach
Every system goes through the forge. Three phases, no shortcuts.
Temper
Map the problem space. Understand constraints, data flows, and failure modes before writing a line of code.
Cast
Build with intent. Clean architecture, tested pipelines, and systems designed to evolve — not just work today.
Sharpen
Harden for production. Load testing, observability, and the edge-case handling that keeps things running at 3 AM.
Got a system that needs building?
I take on a limited number of projects at a time. If you're working on something ambitious, let's talk.
Get in Toucharrow_forward