Data Quality Specialist for US Equity

About the project We build a proprietary investment-research dataset covering ~3,600 U.S. public companies, with point-in-time history from 2008 to the present. The data is tested against equity returns and must hold up to institutional-investor scrutiny: no survivorship bias, no silent history revisions, no entities dropping out because a ticker changed. We generate a set of large files every month (multi-GB files, ~7M company-month rows) and run a structured validation protocol against the prior accepted version. It catches real issues — mapping breaks, coverage drops, silent dropouts — but every month surfaces new ones, and we need someone who owns this end to end. The role: You will be the person who guarantees that every row in this dataset can be trusted — not by eyeballing rows one at a time, but by running checks that touch 100% of the data automatically, then personally chasing every exception to root cause. When a well-covered company silently stops receiving data, you notice, you figure out whether it was a ticker change, a share-class remap, a corporate action, or a vendor bug, and you drive it to a fix. You will: Own the monthly validation cycle: compare each delivery against the prior accepted baseline (coverage, value drift, dropped/added securities, ID-mapping breaks) and issue a written PASS/FAIL verdict before the data goes anywhere. Own the entity-mapping layer: market identifiers (ISIN, CUSIP, ticker, FIGI, CIK, Bloomberg ID) ↔ vendor IDs, including corporate actions (mergers, spin-offs, restructures, renames), multi-share-class structures, and delistings — all point-in-time. Track down every anomaly: keep a living issues log, separate expected behavior from genuine breaks, and write tickets with reproducible evidence — specific companies, dates, and row counts. Harden the tooling: codify numeric tolerances, add automated checks for each new class of issue, and leave a findings report behind every run that a stranger could pick up. You are: Hands-on with U.S. equity historical data: security masters, ID mapping, corporate actions, point-in-time / survivorship-bias-free methodology. Obsessive, in the best sense. A value that moved 0.02 when it should have moved 0.00 bothers you until you know why. Comfortable with Cowork Independent. You'll get context, a working toolkit, and documentation, then you'll run.