Research Scientist, Pre-training Research

present of AI; San Francisco (CA); On-site; USD 150,000 - 230,000; Full time

Job summary

Chai Discovery in San Francisco is seeking a Research Scientist focused on pre-training foundation models for molecular structure and interactions. You will explore architecture choices, training objectives, and data mixes with an emphasis on large-model experiments and biology-driven research.

The role rewards strong math/physics grounding and hands-on experience in training frontier models, with opportunities to connect AI advances to drug discovery initiatives.

Qualifications

• Experience designing and training large models.
• Strong background in generative modelling and scaling laws.
• Solid grounding in computer science, mathematics or physics.

Responsibilities

• Pre-train foundation models for molecular structure and interactions.
• Investigate architecture choices, training objectives and data mixes.
• Demonstrate ability to run large-model training and report results clearly.

Skills

Large-model training

Generative modelling

Architecture

Scaling laws

Computer science

Machine learning

Mathematics

Physics

Job description

Research Scientist, Pre-training Research

AI & Data · San Francisco office · FullTime

The role centers on pre-training foundation models for molecular structure and interaction, with emphasis on architecture choices, training objectives, and data mixes. A strong candidate brings hands-on work designing and training large models, depth in generative modelling, architecture, or scaling laws, and solid grounding in computer science, machine learning, math, or physics. Evidence of getting a model to work or a real contribution to the field matters more than a fixed checklist.

How it fits the AI era

The work is itself AI research, so the tools shift the day to day rather than replace it. Much of the craft is reading training curves, judging when to change phases of a recipe, and simplifying designs, which still depends on human judgment. Adapting ideas from language, image, and video models into biology keeps the role tied to fast moving outside work.

How it builds AI experience

A candidate would gain direct experience training frontier models on biochemical data and learning how data mixes and objectives shape results. Building this means showing prior large model training runs, clear reasoning about scaling, and familiarity with adapting methods from other domains. Interest in molecular design helps connect the research to the company's drug discovery work.

Worth showing

• Strong math or physics

This is our own read, written from the employer's posting and the company's Spotlight. The employer's posting is the source of truth for the role.

About Chai Discovery

Frontier AI foundation models and computer-aided design suite to predict and reprogram biochemical molecules, per the company.