AI Systems Architect · Fractional CTO

I've shipped the AI system
you're trying to build.

Multi-tenant RAG on event-driven AWS — 1,000+ concurrent connections, thousands of requests a second, 99.9% uptime, for 50+ regulated organizations. I design the architecture, build the pipelines and evals, wire the audit trails, and lead the engineers who own it. Not another proof-of-concept.

Walk through the architecture Send me your pipeline; I'll tell you where it breaks

Currently consulting · open to the right permanent CTO role

Live system 99.9% uptime

~0 req/sec live throughput

Systems shipped for

1,000+: concurrent connections
thousands: requests / second
99.9%: uptime
50+: regulated orgs

I design the architecture.

Event-driven on AWS, built to survive real load: 1,000+ concurrent connections and thousands of requests a second at 99.9% uptime. The diagram on this page is the shape of a real production system — request in, retrieval and reasoning across vector and graph stores, model calls fanned out across OpenAI, Bedrock, Azure and Mistral.

I build the pipelines and evals.

RAG that earns trust: chunking and retrieval you can defend, evals that catch regressions before your users do, and audit trails wired in from day one. I took my company through ISO 27001 and SOC 2 in ~3 months — because in regulated markets, compliance is part of the architecture, not an afterthought.

I lead the engineers who own it.

I've built a company from 0 to €1M+ revenue and a team from 0 to 10. At idealo, multi-country SSO I led drove 1.4M new accounts in three months — +700% versus forecast. I hand systems over to teams that keep shipping after I've moved on.

Send me your pipeline. I'll tell you where it breaks.

A short, specific read of your RAG or event-driven setup — retrieval, evals, scaling, and the compliance gaps you'll hit in a regulated market. No slides, no proof-of-concept theatre.

Send me your pipeline Walk the architecture again