### **The role**
You'll **own the AI surface of Elayne end to end:** the models, the pipelines, the evaluation, and the standards for what ships. You'll work directly with the founding team and alongside our senior platform engineer: **no layers, no research org, no six-month roadmap between you and production.**
This seat has real scope. The extraction, provenance, and document-generation systems you build are the product, and as we grow you'll shape the AI team around them. **Meaningful equity** comes with that responsibility.
### **The problem**
Estate settlement runs on documents: wills, deeds, death certificates, bank statements, court forms, each with its own idea of what a name or a date looks like. **Our job is to hold one truth across sources that disagree, and prove where every value came from.**
The outputs get filed with courts and land in the hands of grieving families. A hallucinated value isn't an embarrassing screenshot, it's a rejected petition. Provenance and confidence aren't features here. They're the architecture.
If you've been fine-tuning ranking models or building another chatbot and want your ML to carry real legal weight, this is it.
### **What you'll build**
* LLM-driven extraction from messy, multi-format documents, with sub-document citations, confidence scoring, and human-in-the-loop review where the stakes demand it
* Entity resolution across banks, bureaus, government records, and family-provided documents into a single source-linked estate record
* Document generation: structured estate data into court-ready filings, where a wrong value means weeks of delay
* The evaluation and observability layer that makes all of the above trustworthy: ground truth, regression detection, cost and latency control across model providers
### **Who you are**
* You've shipped LLM systems to production for real users: extraction, RAG, agents, or document intelligence, not demos
* You've worked where correctness had consequences and can talk concretely about how you made model outputs defensible
* Strong in Python; comfortable owning infrastructure (pipelines, vector stores, multi-provider routing) without a platform team behind you
* You've been early somewhere, or you've founded the ML function inside a company
* High-conviction, low-ego: you'd rather find what's right than win the argument
### **The fine print**
**Full-time, in New York City with the rest of the team.** This is high-ownership work, and we're upfront that families and courts don't always operate on business hours.
### **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.
Visa sponsorship is not available for this role.
