Build and scale RL environments and evaluation harnesses for long-horizon legal reasoning. Own pipelines converting court filings into contamination-free benchmarks and RL tasks, integrate with partner model APIs, and collaborate with attorneys to create scorable task formats.

Build and maintain pipelines and web scrapers to create a large case-law dataset, implement metadata extraction (citations, overruling/appeals), ensure scraper reliability, optimize core utilities, expose data via APIs/websockets, and benchmark/evaluate tasks.