23 days ago
Berlin, Germany +3 moreMid Level
Responsibilities
- Design the agentic framework for literature-grounded hypothesis generation, simulation workflows, computational validation, and experimental validation.
- Integrate agents with ML models, simulation engines, databases, and heterogeneous compute backends.
- Build pipelines for agents to plan, schedule, execute, and interpret computational tasks at scale and over long periods.
- Use expert chemistry annotations to improve agent planning, retrieval, and decision-making.
- Create evaluations to measure agent effectiveness.
- Build experimental-design agents using Bayesian optimization, active learning, and related sequential decision-making methods.
- Develop multi-fidelity and multi-objective strategies that balance cost, time, and uncertainty across simulations and physical experiments.
- Close the loop between simulation and physical experiments so outcomes become durable knowledge for future agent reasoning.
- Collaborate with chemists, materials scientists, and the Agent team on core orchestration intelligence.
- Work on customer projects and implement their direct technical needs.
Requirements
- Proficiency in a modern ML ecosystem such as PyTorch or JAX and experience taking ML-driven systems from prototype to production.
- Strong software engineering skills for building scalable production systems, including testing, modular design, and scalable ML operations.
- A PhD or Master’s degree with ideally 4–5 years of industry experience; candidates with a PhD and slightly less industry experience may also be considered.
- Interest in applying AI to scientific and materials-discovery challenges.
- Willingness to learn materials science and experimental chemistry terminology.
- Experience with LLM-assisted programming and a thorough understanding of its strengths and weaknesses.
- Preferred experience applying ML to materials science, chemistry, or drug discovery.
- Preferred experience with agentic frameworks and LLM-powered applications.
- Preferred experience with Bayesian optimization, active learning, bandits, or reinforcement learning in real-world systems.
- Preferred experience with planning models, self-improving systems, multi-tool agents, or RLHF/RLAIF workflows.
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, in addition to local public holidays.
- Paid parental leave of 26 weeks for primary caregivers and 12 weeks for secondary caregivers.
- Professional development budget.
- Office-based work three days per week in Cambridge, London, Amsterdam, or Berlin, with regular travel to other offices potentially required.
- Interdisciplinary collaboration with AI researchers, computational chemists, materials scientists, and experimental scientists.