Staff Machine Learning Engineer

MoonPay is building the operating system for moving value across crypto, stablecoins, tokenized assets, and emerging technologies. With 30M+ customers and 500+ ecosystem partners relying on the platform, the company operates under real regulatory licenses across the US, UK, EU, Canada, and Australia. MoonPay combines high standards with high velocity—this is a place where outcomes matter more than process, and impact drives everything.

We operate in fraud detection and prevention, an adversarial space where attackers constantly evolve and every wrong decision has direct financial consequences. You'll own the real-time decisioning system that powers our transaction platform, from the serving path and feature infrastructure through to the underlying models and the machinery needed to ship changes safely to production.

What you'll do

• Lead technical strategy through ambiguity by translating vague problems into well-scoped solutions, setting the bar through rigorous reviews and clear standards that stick
• Develop feature infrastructure across batch, near-real-time, and in-request serving paths, managing specific freshness budgets for each and keeping training aligned with production behavior
• Own the services that score transactions in-flight within strict latency budgets, designing graceful degradation paths and deciding what happens when models can't respond
• Build and maintain feedback loops that capture every decision and outcome, including counterfactual data from blocked transactions, so the system learns from itself
• Mature the replay, shadow, and staged-rollout tooling until model changes become routine and reversible, then own models across their full lifecycle from training through retirement
• Scale the platform as volume and complexity grow, troubleshooting when offline and production metrics diverge and resolving the root causes

What you'll bring

• Hands-on experience building and operating high-availability services that execute within hard latency budgets on critical paths, with robust fallback mechanisms
• Systems thinking that identifies failure points, manages graceful degradation, and ensures the system remains responsive when dependencies fail
• Engineering craft: code that's tested, typed, and correct when events arrive twice, late, or out of order
• End-to-end ownership of feature or data pipelines in production, including debugging cases where training and serving diverged
• Proven ability to take an unscoped problem and ship it while raising the level of engineers around you

Nice to have

• Experience building systems where explainability matters: audit trails, attribution, or defending model behavior to non-technical audiences
• Work on anomaly detection and identifying novel attack patterns without existing labels
• Familiarity with GCP, BigQuery, Bigtable, Memorystore, Vertex AI, or Kubernetes

What we offer

• Competitive salary package
• Equity ownership so you share in the company's growth
• Hybrid work in London with the team meeting in office roughly 1-2 days per week
• A high-accountability environment where you own real problems and ship continuously