• Rewrote build system to unblock ML research usage of the Research Super Cluster’s storage system

• Implemented flash storage backend for fast and reliable access to 2000PB (2 exabytes) of AI storage

• Created API for large language models and synthesized research for efficiency optimizations

• Built out AWS environment, implementing NAT/VPN PrivateLink to firm networks, batch computation environment, dynamically allocated live & paper trading environments for testing complete stack in “prod” without risk

• Implemented ML-based sentiment models with a $50mm book using Spark

• Implemented caching, distributed, snapshotting filesystem for all production data for stable research

• Created numerous research tools for signal construction and tradeability/factor analysis based on Pandas, Dask, and Spark

• Improved existing technical models by engineering new features and improving training techniques

• Migrated PM team to Bazel, integrating 3 disparate build systems into a unified polyglot build environment

• First senior engineer hire in Boulder site, mentored other engineers and established office culture

• Integrated 7 systems created by teams across 3 other sites to define and launch new attribution product

• Designed and implemented serverless realtime data pipeline on AWS handling 25k transactions per second