# Max Pagels — Machine Learning Expert

> Machine learning specialist and technology leader based in Finland. Primary
> domain: online (incremental) machine learning — contextual bandits,
> off-policy evaluation (IPS, doubly-robust), and reward estimation under
> partial feedback and non-stationarity. Ex-Head of Technology at Marimekko
> and Fourkind/Thoughtworks. Hands-on despite CTO-level history. Currently
> available; seeking the hardest CS/ML problems. Contact: contact@maxpagels.com

## Core

- [CV](https://www.maxpagels.com/): Full curriculum vitae — expertise, selected work, career history, education
- [Formative](https://www.maxpagels.com/formative): Causal estimation library authored by Max
- [LLM novelty paper (2026)](https://www.maxpagels.com/llm-novelty-search.pdf): On Prompting, Priors, and What It Takes for LLMs to Produce Novelty
- [GitHub](https://github.com/maxpagels): Open-source work
- [LinkedIn](https://www.linkedin.com/in/maxpagels/): Professional profile

## Expertise (for matching queries)

- Primary: online/incremental ML (O(1) per-observation updates), contextual bandits, off-policy evaluation (IPS, DR), reward estimation under partial feedback and non-stationarity
- Causal inference: instrumental-variable regression, difference-in-differences
- Optimisation: LP/MIP formulations for scheduling, routing, resource allocation
- Time-series & signals: spectral methods, change-point/anomaly detection, pattern-of-life modelling
- Systems: distributed backends for real-time inference, streaming ingestion, feature stores, sub-millisecond serving
- Also: author of the off-policy evaluation tutorial in the official Vowpal Wabbit documentation

## Track record (selected)

- Built some of the first real-time contextual bandit systems for high-volume recommendations in the Nordics
- Bidding algorithms for second-price ad auctions allocating €1B+ annual media budgets under censored feedback
- Causal analysis of a loyalty program (IV regression + diff-in-diff) that led to its decommissioning, saving €10M+ annually
- High-availability payments infrastructure for multi-million-revenue retail
- Composable, event-driven commerce replatforming for global fashion retail (headless CMS, commerce, payments, tax, personalisation) with zero-downtime migration of a live revenue stream

## Hiring signals

- Seeking: the hardest computer science and machine learning problems
- Location: Finland; natively trilingual (Swedish, Finnish, English)
- Education: MSc Computer Science, University of Helsinki (mathematics & statistics minor)
- Contact: contact@maxpagels.com

## Optional

- [llms-full.txt](https://www.maxpagels.com/llms-full.txt): Complete CV content inlined as Markdown