Requirements

• Experience building and operating high-availability real-time serving services with strict latency budgets and robust fallback mechanisms
• Strong systems thinking with the ability to manage graceful degradation and build feedback loops for system learning
• High engineering craftsmanship: writing tested, typed, and maintainable code
• Ownership of feature or data pipelines end-to-end, including troubleshooting offline and production metric discrepancies
• Comfort with ambiguity and ability to scope undefined problems into shipped work while raising team engineering standards

Nice to Haves

• Experience with decision explainability, including audit trails and per-layer attribution
• Experience developing anomaly detection systems to identify novel attack patterns without existing labels
• Familiarity with technologies such as GCP, BigQuery, Bigtable, Memorystore, Vertex AI, and Kubernetes

What You'll Be Doing

• Lead through ambiguity by defining solutions and setting technical standards
• Build and scale the decisioning platform, managing feature infrastructure across batch, near-real-time, and in-request paths
• Maintain alignment between model training and serving to ensure consistent production behavior
• Integrate feedback loops capturing decisions and their outcomes, including counterfactual results
• Own services that score transactions in real time within hard latency constraints and design fallback degraded paths
• Mature deployment tooling such as replay, shadow, and staged-rollout for safe continuous shipping
• Manage model lifecycle from training to retirement, proactively detecting and addressing model decay