Machine Learning Scientist I - Generative & Agentic AI for Drug Discovery

Frontier Medicines is looking for a Machine Learning Scientist to join our AI team in South San Francisco. You’ll build the AI systems our scientists use every day: LLM-powered agents that work across our research data, generative models that design new covalent molecules, and predictive models trained on one of the largest proprietary covalent chemistry and chemoproteomics datasets in the industry.

This role suits someone with a research mindset who wants their work to reach real decisions. You’ll work in a small team with short communication lines to medicinal chemistry, biology, and computational chemistry. Prior drug discovery experience is not required. If you bring strong AI fundamentals and curiosity, we’ll teach you the chemistry.

• →Build agentic AI systems for scientific workflows, including multi-agent orchestration, tool calling and MCP integrations, task decomposition, and human-in-the-loop checkpoints. These systems connect to our scientific databases and compound registry.
• →Develop and adapt generative models (diffusion, autoregressive transformers, RL-guided generation) to explore new covalent chemical space.
• →Build multimodal models and RAG pipelines that reason over molecular structures, assay data, protein structures, slide decks, reports, and images.
• →Train and benchmark predictive models for property prediction, virtual screening, and SAR. This includes approaches for low-data settings such as transfer, multi-task, and active learning.
• →Deploy models and agents so researchers can use them, and see your work influence active programs.

• MS or PhD in computer science, machine learning, computational chemistry, physics, or a related field, or a BS with 2–3 years of equivalent industry or research experience.
• A track record of research: publications, preprints, thesis work, or substantial open-source contributions in ML.
• Hands-on experience building with LLMs and agents, generative models, or multimodal models.
• Strong Python and PyTorch skills, and comfort with modern AI coding tools.
• The ability to take an open-ended scientific problem and turn it into a working system, independently and collaboratively.
• Clear written and verbal communication.
• Legal authorization to work in the United States.

• Experience with molecular ML, such as GNNs, equivariant models, or molecular representations, or with RDKit and cheminformatics tooling.
• Experience with large-scale or distributed training (DeepSpeed, FSDP, MoE).
• Exposure to biology, chemistry, or drug discovery data.