ML Research Engineer, London
Isomorphic Labs2 months ago
London, United KingdomMid Level / Senior / Staff+
Responsibilities
- Develop and optimize state-of-the-art AI models, robust codebases, data pipelines, and training and evaluation infrastructure.
- Design, implement, and run experiments to evaluate ML model performance and robustness, including data collection and preparation.
- Implement algorithms and software for model analysis, evaluation, inference, and performance optimization.
- Advise on productionizing and integrating AI/ML models, and monitor and refine their behavior.
- Develop specialized tools and frameworks that support model research, evaluation, and deployment.
- Collaborate with research scientists and engineers through discussions, knowledge sharing, and code reviews.
- Identify technical challenges, stay current with AI advances, and deliver independent engineering projects supporting foundation and applied model platforms.
Requirements
- Master’s or PhD in a highly quantitative field such as Computer Science, AI, Physics, or Mathematics, or equivalent practical experience.
- Deep understanding of machine learning principles and techniques.
- Strong proficiency with deep learning frameworks such as JAX or PyTorch.
- Hands-on experience with modern architectures including Transformers, GNNs, and Diffusion Models.
- Experience taking models through the full lifecycle, including scoping, data analysis, training, debugging, evaluation, benchmarking, and deployment.
- Excellent software development skills with strong algorithms and data structures fundamentals.
- Strong collaboration, written communication, verbal communication, self-direction, and ability to navigate ambiguity.
- Preferred: scientific contributions or state-of-the-art AI model work, distributed multi-GPU or multi-node training, performance optimization, computational biology or drug discovery knowledge, applied ML experience, and low-level infrastructure engineering.
Benefits
- Hybrid working requires coming into the London office three days per week, currently Tuesday, Wednesday, and one additional team-dependent day.
- The company supports accommodations for additional needs that affect the hybrid working approach.
Categories