Senior / Staff Machine Learning Scientist
HybridSan Francisco, California, United States
Job Summary
Build generative and foundation chemistry models for molecular design, advancing retrosynthesis by leveraging reaction databases and robot-execution data. Apply computer vision to transform robot telemetry into models that monitor process state and feed back into experimental control. Prototype agentic workflows that orchestrate models, tools, and the platform to close loops between proposal, execution, observation, and learning. Productionise models into a reproducible, API-first toolkit while partnering with Infrastructure on GPU training and HPC. Mentor junior ML scientists and represent Chemify's AI/ML capability externally. Set technical direction across the AI/ML stack to lead cross-cutting initiatives spanning chemistry models, vision, and agents.
Required Qualifications
- PhD or equivalent experience in Machine Learning, Computer Science, Statistics, Physics, or a related quantitative field
- 5+ years (Senior) or 8+ years (Staff) of hands-on applied ML experience, including production-grade work
- Deep familiarity with modern deep learning stack (PyTorch or JAX)
- Breadth across at least two of: generative models (diffusion, autoregressive, flow-based), graph and equivariant networks, vision (CNNs, ViTs, multimodal LLMs), search and planning (MCTS, A*), or agentic / RL systems
- Experience taking ML from prototype to production: reproducible pipelines, distributed jobs, and batch workflows on cloud (AWS / GCP / Azure) or HPC
- Strong scientific computing instincts: clean Python, careful experiment design, leakage-aware splits, and rigorous benchmarks
- Clear communication with non-ML scientists and engineers
- Willingness to pick up new domains (you don't need to know chemistry on day one)
- (Staff level) A track record of technical leadership: mentoring, setting standards, and influencing scientific and technical direction beyond your own projects
- Regular travel to our Glasgow HQ / Chemifarm
Desired Qualifications
- Practical experience with active learning, Bayesian optimisation, conformal prediction, or uncertainty quantification in iterative real-world loops
- Familiarity with retrosynthesis ML, computer-aided synthesis planning (CASP), or reaction-condition / yield prediction
- Working knowledge of how ML fits into a drug-discovery or materials-design workflow
- Familiarity with cheminformatics tooling (e.g. RDKit, OpenEye)
- Willingness to pick up cheminformatics tooling
- MLOps fluency: experiment tracking, data versioning, model serving, and observability of deployed models
- A visible track record in the field — peer-reviewed publications, open-source contributions, or public projects that demonstrate your judgement on real ML problems
Hiring someone like this?
Get your role in front of qualified candidates on Sorce.