CuspAI is the frontier AI company on a mission to solve the breakthrough materials needed to power human progress. While nature took billions of years to perfect molecules, we are harnessing AI to unlock trillion-dollar materials breakthroughs in months, not millennia. Our founding team is the most cited in the world, comprised of world-class researchers in AI, chemistry and engineering.
We are working on some of the hardest and most important challenges including energy, clean water, the future of compute, and carbon capture, and this is just the start of what our 'search engine' for next-generation materials will unlock.
We invite you to be part of a diverse, innovative team at the intersection of AI and materials science, working to create impactful partnerships that drive innovation, scalability, and industry collaboration. This work matters. Your work matters.
We’re on the cusp of the on-demand materials era. Join us.
To scale our scientific intelligence and data curation capabilities, we are looking for three Research Engineers* to train and evaluate the foundation models behind our materials discovery platform: representation learning over crystal structures and properties, LLM fine-tuning and post-training for scientific tasks, and the evaluation and data work that makes those models useful. You will work in a small team where each person owns a model end to end, from data and experiment design through training at scale to integration.
*Note that you would be joining as a ‘Member of Technical Staff’, but the job title above hopefully helps to explain the nature of this role.
You will be responsible for building and training specialised foundation models on terabytes of scientific data at massive scale, fundamentally enhancing our platform's multi-modal capabilities.
The modalities span crystal structures, properties, text and images. Depending on your background you may lead on self-supervised representation learning, LLM fine-tuning and post-training (SFT, RL), or the evaluation suites that decide between them.
Research and Model Training at Scale
Design and run experiments: scaling studies, ablations and evaluations that settle which modelling choices survive.
Lead efforts in parallel training of multi-modal foundation models at scale (e.g. orchestrating parallel training across hundreds of GPUs and handling data in the order of Terabytes).
Apply strong software engineering best practices to ensure the systems are scalable, reliable, and maintainable.
Own the data curation and evaluation design for your models; these outlast the modelling decisions.
Scientific Collaboration & Integration
Partner closely with our Chemists and Materials Scientists to integrate computational and experimental workflows into one seamless optimisation loop.
Be proactive in learning the technical vocabulary of materials science and experimental chemistry to facilitate meaningful interactions with domain experts.
Contribute to our mission by ensuring all system designs align with CuspAI's commitment to sustainability and world-changing challenges.
You are someone who gets excited about the opportunity to enable scientists to work on world changing challenges in this domain, with a personal interest in the potential applications of the technology that CuspAI is building.
Strong software engineering skills and a proven track record of building complex, high-performance systems at scale in a production or industrial environment.
Hands-on experience training or evaluating large models in a distributed setting. For senior candidates this means having led a multi-hundred-GPU training effort; for earlier-career candidates, having been a core contributor to one.
Depth in at least one of: large-scale vision or representation learning, LLM pre-training or post-training, reinforcement learning. We care more about one substantial, scale-y project than about breadth.
A deep interest in the connection between AI and materials science or chemistry, and a willingness to rapidly learn the scientific vocabulary of the discipline.
Master’s or PhD in CS, ML or a related quantitative field, or equivalent experience. We are hiring across levels: from strong early-career research engineers (typically 1 to 3 years at a frontier lab or equivalent) to senior and staff engineers who have led model programmes.
Experience with representation learning and computer vision, e.g. self-supervised or contrastive pre-training, learning embeddings, or building models that pull structured information out of images such as figures, plots and diagrams.
RL for post-training (RLHF, GRPO-style methods) or RL in agentic settings.
Experience at a frontier lab or a team that ships models at scale.
Prior experience applying machine learning models specifically to problems in materials science, chemistry, or molecular informatics.
Exposure to multi-modal training (e.g. text, vision, etc.) and how to combine modalities in a meaningful way.
This role could be based in our Cambridge, London, Amsterdam or Berlin offices, with the expectation of being in the office three days per week. Additionally, there may be regular travel required to our other offices for collaboration and project oversight.
A competitive salary: We value and reward impact and growth
Equity in CuspAI: You have a stake in the success of the company
Time off to stay fresh: 28 days holiday (DE, NL, UK) or 21 days holiday (JP, SG, US), in addition to local public holidays
‘Gold Standard’ parental leave: 26 weeks (primary caregiver) and 12 weeks (secondary caregiver) at full pay - we look after you and your family while we work on the most important materials discovery problems together
Professional development budget: We invest in your career development so you can stay up to date with the latest industry knowledge or add to your skills to increase impact and growth
Solve meaningful problems: See how your work has a direct impact on advancing materials science and solving sustainability and climate-related problems through the creation and application of bleeding-edge SOTA technology and revolutionary techniques
True interdisciplinary teamwork: Be part of a deeply collaborative environment bridging AI research, computational chemistry, and experimental science - work with world-class researchers and engineers who enjoy sharing knowledge and supporting each other
Join us in shaping the future of materials with AI. Together, we can create groundbreaking solutions for a more sustainable world.
CuspAI is an equal opportunities employer committed to building a diverse and inclusive workplace. We do not discriminate on the basis of sex, race, religion or belief, ethnic or national origin, disability, age, citizenship, marital, domestic or civil partnership status, sexual orientation, gender identity, pregnancy or related condition (including breastfeeding), veteran status, or any other basis protected by applicable law.
We actively encourage applications from all backgrounds and value the unique perspectives and contributions that diversity brings to our team.
Please let us know if you require any specific adjustments during or after the interview process. We will do everything we can within reason to accommodate.
Build and own infrastructure for post-training open-weight models, including distributed training and rollouts, agent environments, evaluation tools, and reliable production integration in close partnership with researchers.
Build and scale the infrastructure for post-training open-weight models, including distributed training and rollouts, agent environments, evaluation tools, and reliable production serving in close partnership with researchers.
Build and operate the infrastructure behind legal-model post-training, including distributed training and rollouts, agent environments, evaluation tools, and reliable production serving in close partnership with researchers.
Develop and evaluate reinforcement-learning environments and agentic systems that advance AI capabilities in defensive cybersecurity, translating research into production training runs and collaborating across research and security teams.
Improve the judgment and reliability of production AI security agents by advancing their reasoning, evaluations, model selection, and defenses against attacks such as prompt injection.
