Machine Learning Scientist I/II,… at Lila Sciences · Cambridge

Summary: Lila Sciences is the world’s first scientific superintelligence platform and autonomous lab for life, chemistry, and materials science. As a Machine Learning Scientist focused on decision-making, you will design and implement algorithms to optimize data acquisition and accelerate discovery cycles in materials science and physical sciences. Responsibilities: - Design, implement, and collaboratively productionize algorithms that determine a sequence of experimental choices - Maximize the impact of data acquisition under real-world constraints by integrating uncertainty-aware models with practical data collection strategies - Build Bayesian Optimization pipelines with acquisition functions tailored for diverse, real-world scientific settings - Develop episodic reinforcement learning policies for multi-step planning, including safe exploration, early stopping, and budget-aware strategies - Create multi-fidelity and active-learning workflows that combine diverse, noisy data sources and adaptive sampling methods with real-world constraints - Ensure robust uncertainty quantification and calibration for scientific decision-making - Write reliable, reproducible code and services that scale from offline benchmarking to online, real-world deployment - Communicate findings succinctly to scientific, engineering, and leadership audiences; publish or present impactful results when appropriate Required Qualifications: - Advanced degree (PhD or MS with equivalent research/industry experience) in Computer Science, Applied Math/Statistics, Physics, Materials Science, Chemical Engineering, or related field - Strong foundation in sequential decision‑making: Bayesian Optimization, active learning, contextual bandits, model‑based RL, or Bayesian experimental design - Proficiency in Python and modern ML tooling (e.g., PyTorch/JAX; BoTorch/GPyTorch/Ax or similar); strong software engineering practices Preferred Qualifications: - Background in materials/chemistry or physical‑science experimentation, including autonomous/closed‑loop workflows - Familiarity with scientific simulation (e.g., DFT/MD) and integrating surrogate models with simulators - Open‑source contributions or publications in BO/RL/active learning Required Skills: Bayesian Optimization, Reinforcement Learning, Python Important Skills: Active Learning, Uncertainty Quantification Nice-to-Have Skills: Software Engineering Practices, Scientific Simulation, Open-source Contributions Benefits: Bonus potential, Generous early equity