Member of Technical Staff, Compiler

Overview

Compiled Health is a clinical AI research company. We are building a future where the standard of care is executable: where what medicine knows is what medicine does, at every visit, in every setting. The accumulated, evidence-graded knowledge of medicine still lives in documents built to be read, not executed. We compile that knowledge into decision artifacts that are measurable against the standard, deterministic where they must be, and traceable to their source evidence. Our compiler is live in clinical production today: clinicians in public health networks act on its recommendations every day.

About the Role

Clinical standards arrive as prose, tables, algorithms, footnotes, exceptions, and cross-references whose interactions may only become visible when a real patient reaches an edge case. The compiler must recover the latent program and preserve the evidence for every branch.

Models do their heaviest work here, at compile time. You will build the systems that let them propose clinical logic, measure where they fail, and convert successful outputs into versioned artifacts that can be checked deterministically and inspected by clinicians. The role spans original research and production engineering: a good result both teaches us something and improves the compiler we run.

What You'll Do

• Build the pipeline that converts clinical source material into evidence-bound, executable decision artifacts.

• Design the intermediate representation for populations, applicability, derived facts, temporal conditions, exclusions, precedence, and recommendations.

• Build benchmarks and adversarial cases that measure source fidelity, logical completeness, elicitation quality, and recommendation correctness.

• Turn model failures into hypotheses, experiments, training or inference changes, and better system design.

• Develop deterministic validators using static analysis, constraint solving, formal methods, or other appropriate techniques; make ambiguity, contradiction, and source silence explicit; and carry successful research into production with clinicians and engineers.

Skills and QualificationsMinimum qualifications:

• 3+ years shipping ML or ML-adjacent systems to production.

• A record of exceptional technical work in machine learning, compilers, program synthesis, formal methods, natural-language processing, document intelligence, evaluation, or an adjacent field.

• Strong Python and software-engineering fundamentals, including production proficiency with PyTorch and the Transformers ecosystem; you can turn exploratory work into clean, tested systems.

• Research judgment: you can turn a vague failure into a falsifiable hypothesis, design the experiment, interpret noisy evidence, and decide what follows.