Building the Data Factory: RL Environments, Evals, Atlas and the Engineering Stack

Behind the research results is a platform: tool-calling and computer-use RL environments, rubric and verifier pipelines, the Atlas annotation platform, and benchmarks like FinanceQA, UI-Bench, SpreadsheetBench 2 and Legora's BAR. This episode covers the product surface, the stack named in job postings, and what system-design and eval questions tend to look like.