Senior MLOps Engineer

We are looking for a Senior MLOps Engineer to join AI Enablement and play a central role in ensuring that the AI Accelerator’s models move from development to production reliably and keep performing. This is a hands-on operational role with real stakes. The models you deploy and manage will be used to make decisions about which indications to pursue, in which patient population and against which target. When your systems work well, science moves faster and portfolio decision-making gets better.  

You will take full operational ownership of shipped models, managing deployment, monitoring, retraining and lifecycle end-to-end. You will make sure that the IT-provisioned experiment tracking and model registry systems are used effectively, that training and fine-tuning runs are consistently and correctly logged and that model artefacts are registered with full provenance from data through to prediction. You will work closely with ML engineers at model handover, reviewing documentation and signing off before accepting operational ownership.  

This role is for someone who takes pride in operational excellence and who understands that the AI Accelerator models can only realise their impact on the portfolio if they are deployed and performing reliably in production. 

Key Responsibilities 

• Ensure experiment tracking and model registry systems are used effectively across the AI Accelerator with consistent and correct logging of training and fine-tuning runs and model artefacts registered with full provenance 
• Configure, run and troubleshoot distributed training and fine-tuning jobs, ensuring efficient use of available compute and resolving job-level failures  
• Participate in a structured model handovers with ML engineers, reviewing and signing off documentation before accepting full model operational ownership of shipped models 
• Deploy, monitor and manage model serving endpoints, making technical decisions about serving configurations to meet performance requirements of downstream users 
• Take full operational ownership of models in production, managing monitoring, retraining and lifecycle end-to-end. 
• Uphold MLOps standards and practices across the AI Accelerator, contributing to their evolution based on operational experience and keeping teams current with relevant advances in MLOps tooling 

Required Qualifications 

• MSc in Machine Learning, Computer Science, Software Engineering or a related technical field; PhD preferred or the equivalent industry experience. 
• Solid hands-on experience operating ML training and serving workflows in production environments 
• Experience with distributed training frameworks such as PyTorch Distributed, DeepSpeed, FSDP or Ray Train 
• Experience operating experiment tracking systems and model registry systems such as MLflow, Weights and Biases or equivalent 
• Familiarity with CI/CD tooling for ML workflows e.g. cloud-native pipeline services, GitHub Actions or equivalent 
• Solid understanding of cloud infrastructure for ML (compute, storage, networking) that is sufficient to specify requirements clearly and diagnose infrastructure-related issues 
• Awareness of large model training characteristics including memory footprint, compute scaling and parallelisation strategies 
• Familiarity with infrastructure-as-code tooling such as Terraform or cloud-native equivalents 
• Experience working closely with research and ML engineering teams as a platform operator 
• Experience operating ML infrastructure for large foundation model training  
• Familiarity with biomedical AI workloads, such as training foundation models on large- scale multimodal data