Research Intern, Machine Learning

Description

The company is automating chemistry by combining AI and advanced robotics into a loop: AI systems design experiments, robots run them, and the resulting data improves the next generation of models.

The team is small and deliberately ambitious, and brings together ML engineers, chemists, computer scientists, and hardware engineers. It is hiring a machine learning researcher to help train the next generation of chemistry models.

The company trains models for reaction planning and outcome, input material costs, and mass spectra prediction. Most are state-of-the-art and many use proprietary datasets that are larger and higher quality than those in the literature.

The models are mainly used internally in real workflows, so usage, data, and training stay in a tight feedback loop.

Responsibilities

• You will help train the next generation of chemistry models and make previously inaccessible molecules synthesizable.
• You will cover the whole modeling stack: working with lab staff on data acquisition, training models of many shapes and sizes, and deploying them into experiments.
• You will build and maintain infrastructure for training a varied set of models, including data pipelines and system work such as parallelism and quantization.
• The goal is models with superhuman chemistry intuition and experimental capability.

Requirements

• You have strong machine learning fundamentals and a solid grasp of modern empirical ML practice.
• You have built new training methods or worked on novel tasks or datasets.
• You have worked in a startup, research group, competition team, or similar setting with real ownership and limited resources. This is especially relevant.
• You know PyTorch or another ML framework and have a working knowledge of basic ML theory.
• You enjoy working across the full ML stack: data, training, inference, and deployment.
• You are comfortable crossing disciplines and picking up unfamiliar technical areas.
• You communicate clearly in writing and speaking.
• You are curious about chemistry.
• You prefer building, testing, and learning from real systems.
• No formal education is required. A background in a deeply quantitative field, ideally with ML training experience, is ideal.
• Helpful but optional: experience with LLM training (especially mid- and post-training), ML systems work (scaling, kernels, parallelism), chemistry models such as retrosynthesis or mass spec modeling or cheminformatics, or active learning and low-data techniques. No one is expected to have all of these.
• You can work extended hours and weekends when needed.
• You work fully in person at the South San Francisco lab.
• You can work safely in an active chemistry lab around scientific equipment.
• This role involves no lab work, but some projects need an understanding of lab workflows.

Conditions

The role is fully on-site in the South San Francisco lab.

Hours can extend into evenings and weekends when needed.

This position does not involve lab work, though some projects need familiarity with lab workflows.

Safe working practices in an active chemistry lab are required.