Senior Machine Learning Engineer

Description

The AI Insights team builds the insight layer on top of the company's investment data. The work covers multi-agent systems that answer investment questions with citations, evaluation methods that show whether those answers are sound, agents that carry out expert analyst workflows from start to finish, and extraction that turns research notes, calls and filings into reusable insight.

Some problems are better served by a conventional supervised model, and part of the job is telling which problems suit which approach.

Partners span the investment organization, including research analysts, governance specialists, and the portfolio and order management teams. Each partner has its own data, workflow and view of a good answer. You work deeply with one partner at a time and learn parts of the business most engineers never see.

Whether these systems work, and how to verify it, is still an open question in the field, and answering it here is central to the role. This is applied work with a delivery deadline, not a research lab: answers must hold up for people making real investment decisions and must arrive as working software.

Everyone on the team builds. You will not hand off a design and only review what comes back. A partner engineering team owns the platform, so your time goes to insight and evaluation.

The company manages investments for clients and has more than 9,000 associates across 30+ offices worldwide.

Responsibilities

• Turn a vague question from an investment professional into a defined problem: what is actually being asked, what would count as an answer, and what evidence would settle it.
• Extract signal from messy, incomplete data, and separate real results from leakage, lucky splits and metrics that overstate performance.
• Design evaluations for Generative AI systems: eval sets, success criteria, LLM-as-judge and its failure modes, and judgment about what a metric actually measures.
• Run the experiment that settles a question the team is debating, and document it so others can reproduce it, including the criteria set before seeing results.
• Design and build agent systems: break down the task, choose the orchestration, decide where a human reviews the output, and use a single model call or a plain deterministic step when that is the more sound choice.
• Build prototypes yourself, end to end, using AI coding tools to move quickly while keeping the code clean and working.
• Take projects from rough idea to daily use, starting with a short design shaped together with the team.
• Raise the team's standard through design and code review, and by mentoring others on experimental design and rigor.

Requirements

• A track record of pulling real signal from messy, ambiguous data, with clean evaluations and healthy skepticism toward results that look too good.
• You find the core constraint in an unfamiliar problem without guidance and prefer a reusable structure to a one-off fix.
• You start from the problem and its constraints rather than a favourite tool, and choose the simplest approach that works.
• Production experience in applied ML and Generative AI: real problems taken end to end, from data understanding through evaluation to something people used.
• AI acumen: you learn new tools to understand how they work, use AI coding assistants daily, and can name what you built with them, where they helped and where you had to take over.
• You explain trade-offs clearly to non-technical partners, admit when you do not know something, state opposing views fairly, improve the people around you, and commit to a direction you did not choose.
• Ownership: you have driven ambiguous work to a result on your own.
• Expected: 7+ years of professional experience and still hands-on. You can build a working prototype, read code critically and steer AI coding tools to clean output. This is about building enough to make your research real, not algorithm puzzles.
• Preferred: experience designing and evaluating multi-agent or tool-using systems, with a clear view of where they fail.
• Preferred: experience building evaluation infrastructure, including eval sets, offline and online measurement, and regression and drift detection.
• Preferred: a finance or investment management background, or a proven ability to become fluent in an unfamiliar domain quickly.

Conditions

Much of the week goes to working sessions: brainstorming, design review and pair programming. The problems are genuinely ambiguous, and not all of them succeed.

The team values rigor (trying to break your own result first), ownership, humility (changing your mind when a better argument appears) and pragmatism (knowing when a rough answer is enough). Disagreement stays focused on the work, not the person.

Performance is reviewed annually, and pay is designed to reward the value you provide.

Base salary range for Southern California: $201,683-$322,693. Base salary range for Seattle: $201,683-$322,693.

On top of base salary, you are eligible for an individual annual performance bonus and the company's annual profitability bonus, plus a retirement plan funded by the company at 15% of qualifying pay. Plan guidelines, restrictions and vesting requirements apply.

Temporary US roles are not covered by the pay and benefit plans above.

Under all applicable laws, the company bars unlawful discrimination in hiring and employment decisions, including on the basis of citizenship status.

Benefits

• Starting on your first day, you get generous time away and health coverage.
• Flexible work options are available.
• Two-for-one matching on your charitable gifts, and the chance to secure annual grants for organizations you care about.
• Self-paced learning resources, available on demand, for sharpening current skills and picking up new ones.
• Resources to support your life outside work, whether you are raising a family, volunteering or exploring other career paths.