Senior Machine Learning Operations Engineer

Proximie, a company dedicated to transforming the operating room into a data-rich environment that drives better insights and decision-making. As a Senior Machine Learning Operations Engineer, you will be responsible for owning the platform and infrastructure that deploys our machine learning solutions into live hospitals and ensures their performance. You will work closely with machine learning engineers and scientists to turn their models into production services that meet our latency, cost, and reliability targets. This role requires precision, judgment, and a deep understanding of how machine learning behaves in production at scale.

Implementing monitoring for model performance, data drift, and system health, and ensuring the right people are alerted when something moves.

About Proximie

If you are excited by complex problems with life-changing impact and want to build the systems that get technology like this into the most critical environments in healthcare, we would love to hear from you

Practical MLOps: experiment tracking and model versioning with MLflow, hyperparameter optimisation, and reproducible, configuration-driven training runs

Responsibilities

• Ownership of the platform and infrastructure that carry machine learning solutions out of development and into live hospitals.
• Building and maintaining SageMaker training, processing, and inference workloads, and the pipelines that orchestrate them.

Requirements

• Able to explain technical trade-offs clearly to people who are not engineers
• Production experience on AWS, specifically Amazon SageMaker: training and processing jobs, pipelines, model registry, endpoints
• Expert Python, with deep hands-on PyTorch experience (TensorFlow also welcome) and the surrounding ecosystem: Hugging Face Transformers, timm, scikit-learn, NumPy, pandas, OpenCV
• Strong software engineering fundamentals: Git and pull-request workflow, automated testing, CI/CD

Preferred

• Containerisation and infrastructure as code: Docker and CUDA-based GPU images, plus enough Terraform to ship your own model to an environment through code review rather than a console
• Event-driven and streaming architectures on AWS: Step Functions, Lambda, Kinesis, ECS, DynamoDB
• Exposure to healthcare, medical devices or another regulated industry
• Model monitoring in production: inference logging and drift detection