Our correction record is public. Your job is to keep it short.
ABOUT SPEC RESEARCH
SPEC Research is an independent research house on one mission: telling the world's largest investors which AI claims are real. We publish AIMI, the AI Maturity Index, scoring 27 sovereign wealth funds and 26 public pension funds on the AI they build and apply themselves, and The State of AI Alpha, a report on what AI actually delivers inside investment processes, built on a catalog of 100 scored papers. Every scored fund receives its scorecard and right of reply before the index prints.
RESPONSIBILITIES
- Run the pre-publication verification pass on everything the house prints: AIMI editions, The State of AI Alpha, forward-test results, press materials. Nothing that scores or names a fund goes out until you have traced it to its source.
- Adjudicate right-of-reply disputes in edition season: when a scored fund challenges a number, rule on the evidence and document the ruling.
- Own the public correction record: verify reported errors, publish corrections, add to the record without drama.
- Red-team the house's own methods the way the house red-teams everyone else's: sample the evidence packs behind fund scores, re-run the checks, report what breaks.
- Enforce the sourcing standard: every published claim tagged to a primary source another researcher could audit, with inference never rendered as fact.
REQUIREMENTS
- A background where being verifiably right was the job: fact-checking, audit, manager due diligence, academic peer review, financial journalism, or investment research.
- 3+ years in one of those settings.
- Statistical literacy: you can read a methods section and judge a backtest, an out-of-sample claim, or a regression table without help.
- Working knowledge of current AI and machine learning: you understand what LLMs, supervised models, and production ML systems are, and can tell a deployment from a press release.
- Familiarity with institutional investors and their disclosures: annual reports, mandates, governance structures.
- Excellent written English: plain, precise, and sourced.
- Comfortable telling the principal his number is wrong, in writing, with the source attached.
PREFERRED
- CFA charter or progress toward it.
- Prior verification, fact-checking, or standards-desk experience at a publication, research house, or fund.
- Python or comparable data skills for checking filings at volume.
- Experience using LLM tools in a verification workflow, with judgment about when not to.
- Arabic, French, or a Nordic language, matching funds the index covers.
WHAT WE OFFER
- The final quality gate on research that lands on the desks of the funds it scores.
- Direct work with the principal on every piece.
- Full-time, remote.
HOW TO APPLY
Submit a CV and one paragraph documenting an error, an overclaim, or an unsupported number in any published SPEC Research piece, and how you caught it. If you checked and found nothing, show what you checked.