AI strategy · ML engineering · Independent
I help e-commerce and SME product teams turn AI experiments into production features that are measured and hold up. 16 years in product, including recommender systems at Seznam scale.
Who I work with
For online stores that have tried AI and want it to move store metrics, not slideware.
For 10–250-person teams that want one AI thing in production, done properly.
What I do
Answer questions over your own content and data, with citations.
Conversational front-ends done properly: scoped, evaluated, safe.
Multi-step systems that call tools and take actions, with guardrails.
Recommenders, scoring, forecasting, segmentation: when an LLM is the wrong tool.
Know whether your AI is any good, and catch it when it regresses.
The difference
Most options are either strategists who hand you a deck, or developers who build what they are told. I do the scoping (which use case, LLM-API vs a model, RAG vs a recommender, what ‘good’ means) and the build: prototype, integration, evaluation, handover. One person, one accountable line from idea to a measured result.
Selected work
Discovery and ranking across 50M+ requests/day: clickbait mitigation, long-tail, an experimentation framework.
Co-founded and scaled an email platform from one retailer's internal tool to a SaaS running 100M+ emails/month for 196 brands.
Product for a DMP processing 500M+ user events/day for 50+ media agencies: real-time segmentation, privacy-first.
Named client references available on request.
See all workWriting
About
I started out watching companies sit on data they never used. That led to co-founding an email platform, then a decade at Seznam as senior PM for its recommender systems. I now work solo, scoping and build, with e-commerce and SME product teams, from the EU.
More about how I workThirty minutes, no pitch. Your situation, your data, whether this is even the right problem.