Machine Learning Engineer

About the role

We are a vibrant and intensely mission-driven team in San Francisco, comprising members from MIT, Harvard Medical School, Roche, ETH, and Dana-Farber.

We value speed and rigor, coupled with excitement, drive, and a strong work ethic. In an early-stage environment, we value people who can bring clarity to open-ended problems, take ownership of the next steps, and follow through with energy.

Capable Labs is a place for ambitious, high-integrity people who want to become dramatically better. You will be surrounded by people who care intensely about the work, get close feedback from the people making scientific and company-defining decisions, and have room to own increasingly important problems.

We believe excellent work should be met with meaningful reward, ownership, and trust. Responsibility is earned through contribution, not title alone: anyone who demonstrates the judgment, rigor, and follow-through to move important work forward can earn meaningful scope.

ABOUT THE ROLE

You will build machine-learning systems that remove real bottlenecks from drug discovery and development. The role spans research and engineering: identifying valuable problems, adapting modern biomolecular models, building tools for scientists, and closing the loop between model predictions and wet-lab results. Success is measured by whether the systems accelerate better experiments and better drug-development decisions.

RESPONSIBILITIES

- Work with scientists to identify high-value bottlenecks in drug discovery and development where machine learning can materially improve speed or decision quality.

- Build systems for experiment planning, literature triage, protocol drafting, in silico screening, candidate generation, filtering, and predictive analysis.

- Fine-tune and apply biomolecular models such as ESM, AlphaFold-family models, RFdiffusion, ProteinMPNN, and related approaches using Capable’s data.

- Develop candidate-analysis workflows that may include molecular dynamics, post-training, probing, evaluation, and other fit-for-purpose computational methods.

- Work directly with wet-lab scientists and operators to automate preclinical or clinical-development workflows and make tools usable in practice.

- Build active-learning loops that connect in silico predictions to in vivo results.

- Create internal evaluations that measure whether models and tools improve experimental throughput, candidate quality, or program decisions.

IDEAL QUALIFICATIONS

- Strong research judgment in biomolecular modeling and drug development, or a demonstrated ability and desire to develop that judgment quickly.

- The ability to own an ambiguous problem end to end, from identifying the useful question through building, evaluating, and improving a working system.

- A practical interest in wet-lab reality and in building tools around the constraints of experiments, operators, data quality, and scientific decisions.