AI Engineer
Century Health
- New York, United States
- Hybrid
- Mid Level
- 2+ years
- PhD or doctorate
- Posted
Skills
Required
- ML / AI
- agents
- structured extraction / tool use / fine-tuning
- Python
- LLM-based
- LLM harnesses
- modern AI
Preferred
- post-training / reinforcement learning / RLHF
- clinical or healthcare data
Demand is growing quickly, including a recent multi-year partnership to curate hundreds of thousands of patient records. As one of our first AI hires, you'll get significant ownership early, have the autonomy to propose what we build next, contribute to research we plan to publish, and grow quickly as the company scales, with equity in what you help create. We're also a small, fun team that genuinely enjoys working together.
About Century Health
We're hiring an AI Engineer to help shape how Century Health builds AI for life sciences.
You will work with 6M+ longitudinal patient records that combine physician notes, imaging, and structured data from community practices most of the industry can't reach.
Our first products extract research-grade data from these records and answer partners' questions in plain language.
They are only the start.
You will have real ownership and autonomy: you'll bring ideas for what we build next, take them from first prototype to production, and work on cutting-edge problems alongside our ML, data science, and clinical teams.
The role is a mix of building, customer work, and science, including research we plan to publish.
Compensation: $150-200K base + equity Benefits: Top-tier healthcare, vision, and dental coverage Experience: 2-6 years building ML or AI systems, including recent hands-on production work with LLMs Location: Hybrid, NYC office Reporting to: Founding AI Engineer
Century Health is on a mission to accelerate medical breakthroughs using AI and real-world clinical data.
Today, answering critical clinical and commercial questions is slow because rich clinical data is siloed across health systems, academic institutions, and private practices. Our platform transforms this complex real-world data into research-ready datasets and rapid insights for life sciences partners, while protecting patient privacy and supporting the research community.
We partner with specialty provider practices across the country, which share de-identified patient records with us. Our AI, led by our abstraction model CHARM, reads years of physician notes, imaging reports, and scanned documents and turns them into structured, research-grade variables, with clinical experts in the loop to validate the results. Biopharma companies use these datasets and insights to understand their patients, generate real-world evidence, and get treatments to the people who need them faster.
We are building the infrastructure that turns raw clinical data into impact.
❤️ Patients first
Our work starts and ends with what's best for patients.
🤝 Delight our partners
Our partners place their trust in us, and we take that responsibility seriously. When we're not sure how to do right by them, we ask.
📈 Follow the data
We move fast because trying an idea is the fastest way to prove it works. When we're wrong, we learn and go again.
🏆 Pride in what we build
Making our mark on the world requires ownership and sweating the details. We celebrate doing what we love, doing it well, and doing it together.
🧑🤝🧑 People are the point
We're an AI company, but first we're a group of smart, caring, fun-loving humans who reflect the patients we serve.
We're growing fast and are backed by top-tier investors, so what you build has a good chance to create massive impact for patients! As one of our first AI hires, you will be a key part of the Century Health journey.
Responsibilities
- As AI Engineer you will build, evaluate, and improve the AI systems at the core of our platform. That means post-training and validating models, building and testing evals, fine-tuning LLMs, and writing the code that makes our agents and harnesses better. You will report to our founding AI engineer and work alongside our software engineering, data engineering, data science, and clinical teams, as well as directly with the founders and our business development team.
- The data.
- Our network includes 240+ community clinical sites across neurology, GI, nephrology, ophthalmology, and pulmonology. Each patient record can hold years of physician notes, imaging reports, scanned documents, and structured data, the kind of clinical depth life sciences companies can't find in other real-world datasets.
- The challenge.
- Accuracy. Partners make research and commercial decisions based on what our models extract, so a model has to reason over a patient's history the way a specialist would. CHARM reaches 97% accuracy against clinical expert judgment today, and building evaluations that hold up to physician review across every specialty is still an open problem.
- The opportunity.
- Fine-tune and post-train LLMs, and validate post-trained models before they go into production, including exploring training models for partners.
- Build and test evals for everything we ship, including how we define ground truth with physicians, measure clinical reasoning, and catch errors across a full cohort before a partner does.
- Write code to improve our agent and harness infrastructure.
- Co-develop every part of our AI pipeline, and bring ideas for new capabilities and product components based on what partners need.
- Push the accuracy of our abstraction models on the hardest variables, working with clinicians and annotators across specialties.
- Own abstraction projects for specific provider sites end to end: learn how their clinicians document a condition (for example, an MS relapse), use our in-house tools to abstract the variable start to finish, and find where better tooling could improve the process.
- Design systems that run inference across hundreds of thousands of patient records with reliability, speed, and cost in mind.
- Build the internal tooling and APIs that let our team and AI agents work with our data and model outputs.
- Contribute to research and publications on our methods and results.
- Collaborate closely with the rest of the engineering team, including software engineering and data engineering.
- Communicate clearly what you've built and why, to engineers, clinicians, and partners.
Requirements
- 2-6 years of industry experience building and shipping ML or AI systems; we're not looking for new graduates.
- Strong Python skills, and the ability to ship quickly and make the most of AI coding tools.
- Experience deploying and maintaining LLM-based systems in production.
- Hands-on experience with LLM harnesses and agents, such as structured extraction, tool use, or fine-tuning.
- You stay current with modern AI: you read papers, track new model releases, and bring new techniques into your work.
- Experience designing evaluations for model behavior where ground truth is hard to define.
- Strong engineering fundamentals; you develop and monitor accurate AI pipelines.
- High ownership and autonomy; you're excited to own projects end to end and propose what to build next.
- Strong, low-ego communicator; you can explain a model's errors in plain language and turn feedback from clinicians and nontechnical teammates into a better model.
- Able to work hybrid from our NYC office.
Preferred
- Experience with post-training, reinforcement learning, or RLHF.
- Experience with clinical or healthcare data. If you're new to the domain, you'll learn it quickly alongside our physicians.
- Research experience, such as a PhD or publications in ML.
- These qualifications are guidelines. If you are excited about the mission and believe you would thrive in this role, we encourage you to apply.
- We dig into the data to understand why, and we act on what it tells us. When opinions differ, the data decides.
- ⚡ Bias to action
If you're interested, please apply and we'll be in touch.
