# ML Research Engineer (Machine Learning Force Fields)
## About the Role
You will advance CuspAI's molecular simulation capabilities by developing next-generation computational methods and robust infrastructure. You'll shape the simulation infrastructure that enables evaluation of novel material candidates through atomistic physics, bringing simulations to the accuracy and performance needed for large-scale search campaigns.
## Key Responsibilities
### Models
- Train, fine-tune, and distill machine learning force fields
- Research and develop novel ML force field architectures suited to production simulation workloads
### Systems & Infrastructure
- Integrate models into public and in-house high-performance simulators
- Develop training and inference architectures for large-scale training, data generation, and simulation
- Distribute workloads via Ray to scale across compute infrastructure
- Build modular systems with reusable components across chemistry applications
### Science & Collaboration
- Build an active learning system closing the loop between simulation, data generation, and training
- Develop interfaces for domain scientists to use and extend the system
- Collaborate with computational chemists on DFT data generation and validation
## Requirements
- Motivated by building foundational tools enabling world-changing research
- Demonstrated technical excellence in both research and implementation with track record of high-quality, performant systems
- Exceptional coding skills with strong modern software engineering practices
- Deep production or research experience with distributed machine learning systems
- PhD or comparable professional experience in relevant quantitative field (Computer Science, Physics, Applied Mathematics, Computational Science, Machine Learning) with strong computational methods foundation
- Genuine and explicit interest in AI applications within materials science and chemistry
## Bonus Skills
- Experience deploying, training, and modifying machine learning force fields
- Management of atomistic data
- Density Functional Theory experience
- Molecular simulation methods (MCMC, MD)
- Graph neural network design
- Cloud infrastructure and Kubernetes
- Published research at top-tier ML (NeurIPS, ICML) or computational physics venues
## Location & Schedule
Based in Cambridge, London, Amsterdam, or Berlin offices with expectation of three days per week in office. Regular travel may be required.