Senior AI Engineer

About the role

We’re looking for a Senior AI Engineer to build the intelligence layer behind Elayne.

Estate settlement is an unusually hard AI problem. The truth about an estate is scattered across wills, tax records, deeds, bank statements, court filings, credit data, government databases, and documents that frequently disagree with each other. We need systems that can understand all of it, connect what belongs together, know when they might be wrong, and show exactly where every answer came from.

You’ll own that problem end to end: document intelligence, entity resolution, agentic workflows, evaluation, provenance, and the infrastructure that makes probabilistic systems reliable enough for consequential financial and legal work.

The challenge isn’t getting an LLM to produce an answer. It’s building a system we can trust when the answer matters.

You’ll work directly with the CEO and Head of Engineering, with enormous freedom to determine how we solve these problems and what the AI architecture of Elayne becomes.

What you’ll build

• An AI system that understands an estate: extract and reconcile people, accounts, assets, liabilities, relationships, and legal facts across thousands of messy documents and external data sources
• Agentic financial discovery: build systems that reason across fragmented financial and government data to identify what someone owned, what may be missing, and what should happen next
• AI that knows when it might be wrong: build confidence, provenance, citations, validation, and human review into the architecture so consequential outputs are defensible
• The evaluation system behind everything: create ground truth, regression testing, observability, and model evaluation that let us change prompts, models, and pipelines without quietly making the product worse
• The infrastructure to make it production-grade: own model routing, latency, cost, retries, failure modes, and the pipelines that turn AI capability into a reliable product

This isn’t another chatbot or AI wrapper. A hallucination here can mean a missed account, an incorrect court filing, or an asset that never reaches the person who should inherit it. The systems you build will make sense of decades of someone’s financial life and help transfer what they built to the people they left behind. Correctness and provenance aren’t guardrails around the AI. They’re the foundation of a new category of software for the largest wealth transfer in history.

Who we’re looking for

• You’ve built and operated AI systems in production. You’ve owned document intelligence, extraction, agents, retrieval, or other model-driven systems end to end, including evaluation and reliability
• You know reliability is the hard part. You can talk concretely about evals, regressions, hallucinations, provenance, human escalation, cost blowups, latency, and what broke after you shipped
• You’re an engineer first. You’re exceptional in Python and can architect the pipelines, data systems, APIs, and infrastructure around the models without waiting for another team
• You want the messy problems. Ambiguous documents, conflicting sources, entity resolution, imperfect models, and systems where the ground truth itself can be difficult to establish are interesting to you
• You want ownership, not a research sandbox. You move quickly, care about what reaches production, have strong opinions without needing to win the argument, and want to be accountable for whether the system actually works

How we work

We’re small, fast, and close to the people using what we build. Product decisions can start with a family conversation, a rejected filing, an institutional edge case, or something we had to do manually that software should absorb next.

Everyone here owns outcomes, not tickets. We communicate directly, move quickly, and care deeply about impact and getting the details right.

Location & compensation

NoMad, NYC office

$250,000 to $300,000 + meaningful equity

Medical, dental, and vision insurance + 401(k)

Equal opportunity

Elayne is an equal opportunity employer. We evaluate qualified applicants based on ability, experience, and potential, and do not discriminate on the basis of any protected characteristic under applicable federal, state, or local laws.

Interview Process

• Intro conversation: get to know each other, the role, and what we’re building
• Technical interviews: ML systems design session and project deep dive
• Office session: a working session on a live problem and time with the team
• Offer: we check references, then move fast