Software Engineer (Scheduler), London
Isomorphic Labs1 day ago
London, United KingdomSenior
Responsibilities
- Design, operate, and extend scheduling systems for dynamic batch inference and long-running machine learning training workloads.
- Implement scheduling policies including gang scheduling, topology-aware placement, and preemption to optimize GPU and TPU utilization.
- Collaborate with ML researchers and infrastructure teams to understand workload behavior, troubleshoot bottlenecks, and translate research requirements into platform capabilities.
- Build reliable scheduling systems with clear failure semantics, operational processes, runbooks, and SLOs.
- Build and maintain custom Kubernetes controllers, operators, and CRDs for ML pipeline requirements.
Requirements
- Practical experience building or operating cluster schedulers, batch systems, or queuing frameworks such as Kueue, Volcano, or Slurm.
- Deep hands-on Kubernetes experience, including maintaining custom controllers or operators and working with cluster orchestration APIs and the Kubernetes scheduler.
- Strong practical experience with cloud providers, compute provisioning, networking, and infrastructure as code.
- A solid foundation in DevOps and SRE practices applied to AI accelerator infrastructure.
- Strong programming skills with high proficiency in Python or Go.
- Preferred: experience operating Kueue; familiarity with Nvidia GPUs, Google TPUs, gang scheduling, or collective networking; high-performance computing experience; GCP cloud compute and infrastructure-as-code experience; or ML software engineering or infrastructure SRE experience supporting machine learning research.
Benefits
- Hybrid work requiring attendance in the office three days per week, currently Tuesday, Wednesday, and one additional team-dependent day.
- Equal employment opportunities and accommodations for disabilities or additional needs are provided.
Tech Stack
Categories
About Isomorphic Labs
Isomorphic Labs develops AI models and a computational drug design platform for pharmaceutical R&D, working primarily through discovery collaborations and licensing. Founded in 2021 by Alphabet and headquartered in London, the company builds on advances such as AlphaFold to predict and design novel molecules. It has announced multi-year partnerships with Eli Lilly and Novartis to apply its technology to new therapeutics.