Platform Engineer hos RaySearch i Stockholm

The work spans:

Data curation and management: ingesting, organizing and validating large clinical datasets and medical images, with ETL, databases and storage that keep track of where data came from and how it was used.

Model training: reproducible training pipelines orchestrated with Flyte, for dose prediction, segmentation, detection, classification and image registration, with monitoring of training runs.

Validation: automated evaluation and reporting that connects data, method, code and results, including web-based tools for reviewing images, contours, radiotherapy plans and validation outcomes.

Deployment and ML integration: getting models into production and integrated into our software products, including inference optimization.

Operations: monitoring training and evaluation runs, keeping services and databases up to date, and troubleshooting when things go wrong.

The platform is used by multiple teams and projects, so good design matters: well-designed APIs, solid documentation and reliable services. You will own services from design to operation and work closely with ML engineers, software developers, infrastructure teams, medical physicists and product teams.

Requirements

M.Sc. in Computer Science, Software Engineering, Machine Learning or equivalent experience, and at least 5 years of professional software engineering.

Strong Python skills and experience designing backend services and APIs.

Solid knowledge of relational databases and data modeling, such as PostgreSQL.

Experience with containers, for example Docker, and with designing, running and troubleshooting production services.

Good engineering practices: automated testing, code review, CI/CD and observability.

Experience contributing to architecture decisions in large, shared codebases.

Excellent spoken and written English.

Meriting

• Experience in one or more of the following areas is considered an advantage:

ML platforms and MLOps tooling, such as MLflow and experiment tracking.

PyTorch, and model deployment or inference optimization, such as TensorRT.

Workflow orchestration and ETL. We use Flyte, but experience with Airflow, Prefect, Dagster or similar is just as relevant.

Kubernetes, cloud platforms and infrastructure as code.

Designing platforms and APIs used by many teams or users, including access control and security.

Operating production services with high reliability requirements.

Web application development.

Medical imaging (DICOM, RT structures), healthcare technology or other regulated, safety-critical environments.