Equities Analyst, AI Evaluation

Cobalt is hiring experienced finance professionals to build and review the cases used to test whether frontier AI models can carry out real investment research and financial modeling work. Each case is built around a single company, investment decision, or transaction, and recreates the full body of work an analyst would actually do: the source materials, the intermediate working papers, the successive model versions, and the final deliverable. Your job is to make each case realistic, difficult, and gradable, so that it measures whether a model can do the work at the standard of a strong professional. If you have built models and written memos that real capital decisions depended on, this role puts that experience to work at the frontier of AI evaluation.

What you will do

• Build cases around a single company or transaction, assembling the complete set of materials an analyst would use, such as multi-year annual reports, quarterly reports and footnotes, earnings call transcripts, investor presentations, management accounts, general ledgers or revenue detail, credit agreements, and transaction documents. Each case typically includes at least 15 relevant files.
• Include the intermediate work that shows how the analysis developed, such as due diligence questions and answers, adjustment lists, working papers, successive model versions with revision logs, and the final model, valuation report, or investment or credit memo.
• Write difficult investment research questions tied to a specific as-of date, so that a correct answer depends only on information available at that point in time.
• Write detailed grading rubrics that distinguish an excellent answer from a plausible but flawed one, and that produce consistent scores when applied by different reviewers.
• Write expert hints that point toward the right line of analysis without giving away the answer, so that model performance can be compared with and without guidance.
• Review how frontier AI models attempted each case, both with and without hints, and judge whether a failure reflects a real gap in capability or instead reflects an ambiguous question, missing materials, or an unfair rubric.
• Check that every file opens correctly, with no truncation, missing pages, or encoding errors, and that scanned documents are clearly legible.
• Where materials are de-identified, make sure the original formatting and the differences between model versions are preserved.

Who we are looking for

• At least six years of experience in investment banking, private equity, equity research, credit analysis, hedge funds, corporate development, or transaction advisory.
• Deep, hands‑on expertise building complex financial models in Excel, including three‑statement models, discounted cash flow valuations, leveraged buyout models, and merger models.
• Experience writing investment memos, credit memos, or valuation reports that informed real decisions.
• The judgment to tell a defensible analysis from one that only looks right, and to explain exactly where the difference lies.
• Careful, direct written communication. Much of the job is writing precise questions, rubrics, and hints.

Strong advantages

• A Chartered Financial Analyst charter, a Master of Business Administration, or an accounting qualification such as a Certified Public Accountant license.
• Experience training or reviewing junior analysts, including reviewing models and marking up their work.

Why Cobalt AI

• Work on frontier AI evaluation. Your judgment directly determines which cases are used to measure what leading models can and cannot do in finance.
• See how frontier models are trained and assessed. Reviewing model attempts gives you a close view of where current models succeed in financial analysis and where they break.
• Fit the work around your other commitments. Set your own schedule, with flexible weeks of 10 to 40 hours.
• Shape the work. Strong experts can lead case design, run projects, and help define quality standards.