About

I'm the architect who also builds. When a client tells me "we need feature X", I ask first: Does X actually solve the problem? Only when the requirements are right do I write the first line of code.

On one project, the team gave me a nickname: Worry Eater. Because I grab the problems nobody wants — performance bugs, legacy code, edge cases in production. Not out of masochism, but because that's where the biggest leverage is.

The result: Queries from minutes down to under a second. Not by throwing more hardware at it — by understanding what the database actually does.

In AI projects, it's not the demo that matters — it's the behavior after the third prompt change. That's why I build evaluation from day one.

My assumption on a RAG project: "Vector embeddings are the standard approach." So I built both variants and compared them. The result: LLM-based metadata analysis was 29% better. And GPT-4 beat GPT-5 — lower cost, higher accuracy.

I don't know this because I read it somewhere — I know it because I measured it.

I think along, challenge, and build. I have plenty of leadership experience (4+ years as CTO), but permanent management roles without code? Not my thing.

What I offer: Someone who develops the technical vision AND implements it. Who makes product decisions AND ships them. Who can coordinate a team (when needed), but primarily delivers through results — not through meetings.

For larger projects, I work with Mit Freuden — a network of four experienced software developers, each with 15–20+ years of experience.

Four developers. One direct line. No account managers, no hierarchies, no telephone game. Each of us focuses on one project at a time.

Product Development:

• Holistic thinking: requirements, architecture, UX, business impact
• Direct communication instead of ticket ping-pong
• Technical decisions with business impact in mind

AI & Agent-Native Architecture:

• LLM integration with measurable quality
• Evaluation-first development
• Multi-provider architectures (OpenAI, Anthropic, Google)
• Agent-native systems — software built for agents

Full-Stack Engineering:

• Production systems with billions of records
• Performance optimization (from minutes to under a second)
• Strict TDD — hundreds of specs as a safety net