ML Engineer

WithCoverage•New York, NY

2d•$180,000 - $275,000•Onsite

About The Position

We're looking for an ML Engineer to build the models and data systems that make our AI platform smarter over time. While our AI Agent Engineers focus on orchestrating LLMs into production workflows, you'll go deeper—building the custom models, evaluation systems, and data infrastructure that give our agents (and our business) capabilities that off-the-shelf models can't provide. This is a high-ownership role. You'll work across the full ML lifecycle: identifying where custom models can outperform general-purpose LLMs, building the data pipelines to make that possible, and shipping models that run reliably in production. You'll turn the proprietary data we generate every day—across policies, claims, emails, and client interactions—into a compounding advantage. We're not handing you a roadmap. The surface area for ML at WithCoverage is massive—and we've barely scratched it. We're looking for someone who can look at our data, our domain, and the state of the field—and move faster than the roadmap. This role is based in our NYC office.

Requirements

3+ years building and deploying ML systems in production environments.
Strong experience with NLP: transformers, embeddings, fine-tuning, text classification, information extraction.
You've built ML pipelines end-to-end—data processing, training, evaluation, deployment, monitoring.
Strong fundamentals in software engineering: you write clean code, design sensible systems, and ship consistently.
You think from first principles. You can navigate ambiguity, make tradeoff decisions, and figure out what to build when there's no playbook.
You want ownership. You're excited by autonomy and accountability, not layers of process.

Nice To Haves

Experience with embeddings, vector databases, and retrieval systems is a strong plus.
Comfortable with Python and ML tooling (PyTorch, HuggingFace, etc.). Experience with our broader stack (Node.js, GraphQL, Postgres) is a plus.

Responsibilities

Build Custom Models. Train and deploy models tailored to the insurance domain—document understanding, classification, extraction, risk prediction, and beyond. Identify where fine-tuned or purpose-built models can meaningfully outperform general-purpose LLMs, and build them.
Own Data Infrastructure. Build the pipelines, labeling workflows, and data systems that make ML possible at scale. Turn messy, unstructured insurance data into clean, usable datasets. Design systems for continuous data collection and model improvement.
Build Evaluation & Quality Systems. Design evaluation frameworks that measure model and agent performance with rigor. Build benchmarks, catch regressions, and create the feedback loops that let us iterate with confidence. Make quality measurable, not anecdotal.
Embedded Problem Discovery. Dig into how our business actually works. Understand where predictions could replace guesswork, where classification could replace manual review, where ML could unlock capabilities we don't have today. Identify high-leverage opportunities that nobody has asked for yet. Prioritize ruthlessly. Build what matters most.

Benefits

Competitive compensation that may include equity
Flexible paid time off
Comprehensive benefit plans for medical, dental, vision, life, and disability
Flexible Spending Accounts (FSAs): Health Care FSA and Dependent Care FSA
Commuter Savings Account
Human Interest: 401(k) provider
Time Off: Sick Leave, Family and Medical Leave, Flexible Time Off
Paid Holidays: Observance of all major national holidays
A curated in-office employee experience, designed to foster community, team connections, and innovation, that also includes catered lunches in the office on Fridays for in-office workers
Collaborative, transparent, and fun culture