Research Scientist, Model Training → NovaAtom structure prediction Protein design models RL post-training Menlo Park

Geodesiclab; Menlo Park (CA); Hybrid; USD 120,000 - 210,000; Full time

Job summary

Geodesiclab is seeking a machine learning-focused engineer to develop and train models powering NovaAtom and the company’s protein design systems. You will work on structure prediction architectures and reinforcement-learning loops that translate experimental feedback into better models.

The role involves training NovaAtom-family models, building protein design models for NovaDDE, and designing RL post-training pipelines that weave wet-lab feedback into ongoing model updates.

Qualifications

• Strong background in machine learning with hands-on experience training large models.
• Familiarity with structural biology or protein modeling is a plus, not a requirement.
• Comfortable moving between research and production-quality training infrastructure.

Responsibilities

• Train NovaAtom-family structure prediction models.
• Build and improve protein design models used across NovaDDE.
• Design RL post-training pipelines that incorporate wet-lab feedback into model updates.

Skills

Machine learning

Production ML

Large model training

Job description

About the role

You'll develop and train the models that power NovaAtom and Geodesic's protein design systems — from structure prediction architectures to the reinforcement-learning loops that turn experimental feedback into better models.

What you'll work on

• Training and fine-tuning NovaAtom-family structure prediction models
• Building and improving protein design models used across NovaDDE
• Designing RL post-training pipelines that incorporate wet-lab feedback into model updates

What we're looking for

• Strong background in machine learning, with hands-on experience training large models
• Familiarity with structural biology or protein modeling is a plus, not a requirement
• Comfortable moving between research and production-quality training infrastructure