2 days ago
Berlin, Germany +3 moreEntry Level / Mid Level / Senior / Staff+
Responsibilities
- Design and run scaling studies, ablations, experiments, and evaluations for foundation models.
- Lead or contribute to parallel training of multimodal models across hundreds of GPUs and terabytes of scientific data.
- Build scalable, reliable, and maintainable high-performance software systems.
- Own model data curation and evaluation design.
- Train and evaluate models involving crystal structures, properties, text, and images.
- Collaborate with chemists and materials scientists to integrate computational and experimental workflows.
- Learn and apply materials science and experimental chemistry terminology in cross-disciplinary work.
Requirements
- Strong software engineering skills and a record of building complex, high-performance systems at production or industrial scale.
- Hands-on experience training or evaluating large models in a distributed setting; senior candidates should have led multi-hundred-GPU training efforts, while earlier-career candidates should have been core contributors.
- Depth in at least one area such as large-scale vision or representation learning, LLM pre-training or post-training, or reinforcement learning.
- Master's or PhD in computer science, machine learning, or a related quantitative field, or equivalent experience.
- Experience with representation learning and computer vision, including self-supervised or contrastive pre-training, embeddings, or extracting structured information from images.
- Typically 1 to 3 years at a frontier lab or equivalent for strong early-career candidates, with senior and staff-level candidates also considered.
- Interest in AI applications for materials science or chemistry and willingness to learn the field's technical vocabulary.
- Preferred qualifications include RLHF, GRPO-style methods, reinforcement learning in agentic settings, frontier-lab experience, materials or chemistry applications of machine learning, and multimodal training experience.
Benefits
- Competitive salary and equity in CuspAI.
- 28 days of holiday in Germany, the Netherlands, and the UK, or 21 days in Japan, Singapore, and the US, plus local public holidays.
- 26 weeks of fully paid primary-caregiver parental leave and 12 weeks of fully paid secondary-caregiver parental leave.
- Professional development budget.
- Based in the Cambridge, London, Amsterdam, or Berlin office with an expectation of three days per week in the office.
- Regular travel to other offices may be required for collaboration and project oversight.
Categories
AI ResearchML Engineering
About CuspAI
CuspAI builds an AI-driven materials discovery platform that helps R&D teams in chemicals, energy, and manufacturing design and evaluate novel molecules and materials. The company offers software and models accessed via cloud APIs and collaborations to prioritize candidates, guide experiments, and shorten lab cycles. Founded in 2024 and headquartered in Cambridge, it is a privately held startup focused on applying generative and physics-informed AI to sustainability and next‑generation materials problems.