Working on the ML Platform team to build out the tools and infrastructure to deploy models at scale.
o Architected an organization-wide agentic platform in partnership with Product, standardizing autonomous workflows through a centralized semantic model repository, evaluation framework, shared context layer, and unified serving layer
o Led a two-quarter, cross-org initiative to deliver minute-level feature streaming for online models, processing 30K+ Kafka events/sec with Spark into Iceberg (offline) and DynamoDB (online), and serving up to 1,500 features via Ray Serve at sub-200ms P99 latency
o Directed a build-vs-buy decision for ML monitoring and drove adoption of Arize for large-scale feature and model drift detection, surfacing retraining signals that improved model reliability, trust, and performance
o Built and hardened an ML deployment framework with a unified SDK layer across Metaflow, Ray, Spark, and Feature Store, accelerating experimentation cycles and enabling engineers to self-service model deployment
o Mentored junior engineers and drove engineering excellence by defining Claude Code guidelines, leading senior+ design interviews, and shaping roadmaps through collaboration with ML and product teams
