Machine Learning Engineer - Foundation Models Biology

Machine Learning Engineer - Foundation Models for Biology

Prima Mente – Full‑time, Negotiable – Primarily based in San Francisco or London.

Prima Mente is a frontier biology AI lab that builds general‑purpose biological foundation models, translating discoveries into research and clinical outcomes. Our mission is to deeply understand and protect the brain, while enhancing neurological health. The team comprises AI researchers, experimentalists, clinicians, and operators across London, San Francisco and Dubai.

You will design, implement, and scale foundational AI models and infrastructure for multi‑omics data at massive scale, directly contributing to breakthroughs in scientific understanding and transformative medical applications.

• Implement high‑performance ML algorithms optimized for massive scale, ensuring reliability, efficiency, and scalability.
• Design, develop, and maintain robust experimentation pipelines that enable rapid iteration, precise evaluation, and reproducible research.
• Refactor prototype research code into clean, maintainable, and production‑grade repositories.
• Create high‑speed data‑processing workflows that efficiently handle large‑scale datasets to accelerate experimentation and model development.
• Conduct experimental design, prioritising high‑impact experiments with the best signal‑to‑noise ratio.

• 1 month: Run initial experiments with state‑of‑the‑art ML models, review and implement cutting‑edge research papers, and optimize existing code for efficiency and accuracy.
• 3 months: Own and deliver a prototype model architecture, demonstrate significant algorithmic improvements, and contribute scaled methods for large‑scale data ingestion and training.
• 6 months: Deliver a high‑performance foundation model with key algorithmic optimizations that boost scalability and throughput, and publish internal benchmarks demonstrating significant impact.

You want to redefine what’s possible at the frontier of AI and biology. You’re not required to tick every box; strong applicants often have depth in some areas and a willingness to grow in others.

• Deep understanding of state‑of‑the‑art machine‑learning methodologies and proven experience translating them into practical solutions.
• Solid foundation in distributed computing principles, parallel processing, and algorithmic efficiency.
• Experience optimizing ML algorithms for performance, memory efficiency, and compute resource utilization.
• Deep expertise in modern ML frameworks and tools (PyTorch, JAX, Tensor Flow).
• Familiarity with training, optimizing, and deploying large‑scale models (7B+ parameters) and inference workflows.
• Skilled in designing and implementing scalable data pipelines for rapid ingestion, transformation, and processing.
• Skilled in communicating complex ideas, explaining why particular approaches succeed or fail, and providing insightful critical analyses.
• Low‑level algorithm optimisation: quantization (8‑bit or lower), JIT compilation, XLA/Mosaic/Triton/CUDA.
• Hardware optimisation: GPU/TPU/HPU.
• Fine tuning optimisation (QLora, QDora).
• Large‑scale data processing (above 2T tokens).

Based onsite in San Francisco, US or London, UK. We support O1 (US) and GTV (UK) visa applications.