cost cut at Expat Money
on WCP V5
I build the systems that work while you don't.
Applied AI reliability — production RAG & LLM apps, LangChain agents, evals, FastAPI, deterministic scaffolding.
Now: Founding Engineer at Zap Viagens — privacy-first, self-hosted AI products on open-source models.
Proof hallway
Same person on every surface. Start where you landed.
What this proves
Three and a half years of production AI
Created and led the AI function at Expat Money (2022–2025) — assistants, content pipelines, RAG, and automation running inside a real business. Not a demo.
Everything public, built in the open
Every repo is public. Every project page links to source. You can inspect the code, the tests, the architecture — nothing is hidden behind a private org.
Systems thinking, not feature work
Five-service architectures, agent frameworks, observability SDKs, RAG evaluation harnesses — the repos form a coherent system, not a scattered portfolio.
Shipped proof
Current public repos to inspect first.
WCP-Compliance-Agent-V5
Production-ready WH-347 compliance platform: deterministic validation decides, the LLM explains, and every decision is traceable.
Project pageGitHubgroundtruth
Canonical production RAG platform: hybrid search, citations, refusal logic. Offline-first demo; PostgreSQL/pgvector, Redis, model-provider, and document-parser integrations are optional, not a hosted production deployment.
Project pageGitHubaria-agent
Canonical clean-break agent harness: approval gates, execution tracing, and progressive-disclosure Agent Skills.
Project pageGitHubevalforge
Regression testing for RAG and agentic AI: retrieval and conversational evaluations.
Project pageGitHubagenttrace
Agent observability and replay SDK with cost attribution, prompt-cost reporting, and an absorbed safety-bounded issue→draft-PR workflow.
Project pageGitHubllm-gateway
LLM proxy with routing, guardrails, cost control, and fallback across providers.
Project pageGitHubWriting
Long-form analysis, published.
Taste Is All You Need
Agent-harness economics: model intelligence is commoditizing, and the scarce complement is engineering judgment encoded into the environment around the model — why scaffolding beats model spend.
Read on X