Lead ML Ops Developer in Salford

What you’ll be doing:

• Design and maintain a production MLOps platform on Amazon SageMaker (Studio, Training, Pipelines, Endpoints) — including model registry, automated retraining, drift monitoring, and governance gates.
• Lead the migration of a 12-model production suite (e.g., the CVM suite) from legacy infrastructure to SageMaker, owning parity testing methodology and sign-off.
• Build and maintain CI/CD pipelines (CodePipeline/CodeBuild or equivalent) for automated model promotion across environments.
• Define and enforce IAM least-privilege policies, KMS key management, and VPC/PrivateLink network controls for all ML workloads.
• Create the 'golden template' MLOps patterns — model packaging, versioning, monitoring, and compliance gates — that other teams self-serve from.
• Produce technical documentation and runbooks that enable data science teams to operate pipelines without central bottlenecks.
• Communicate parity gaps, governance trade-offs, and migration risk clearly to non‑technical stakeholders and project sponsors.
• Size and sequence interdependent migration work, making sound technical decisions before all edge cases are known and adapting as issues surface.

AWS & SageMaker (must have):

• Amazon SageMaker (Studio, Training, Pipelines, Endpoints) — expert level; you can architect and operate the full lifecycle.
• AWS IAM — advanced; writes least‑privilege policies from scratch, not just modifies examples.
• Amazon S3 — advanced; including lifecycle policies, encryption, and bucket policies.
• AWS KMS — working knowledge of key management in an ML context.
• CI/CD tooling (CodePipeline / CodeBuild or equivalent) — advanced; you've automated model promotion across environments.

General and technical:

• Python / PySpark — expert; production‑quality code, not just notebook scripts.
• Statistical / parity testing methodology — advanced; you can design and execute parity sign‑off on migrated models.
• MLOps pattern design (model registries, monitoring, governance gates) — expert; you've built and owned these patterns in production.
• Git / version control — advanced; branching strategies, PR workflows, and release tagging for ML artifacts.
• Track record of technical ownership — accountable for platforms that other teams depend on, not just your own workstream.
• Enablement mindset — you build patterns and hand them off so teams self‑serve, rather than becoming a single point of failure.
• Risk communication — able to explain parity gaps, governance trade‑offs, and migration risk to non‑technical audiences.
• Decision‑making under ambiguity — comfortable setting the technical pattern before all edge cases are known and iterating as issues emerge.

Nice to have:

• AWS Step Functions / Lambda for workflow orchestration.
• Amazon CloudWatch / CloudTrail for platform observability and audit.
• AWS Glue / EMR for data processing pipelines.
• AWS Lake Formation and SageMaker Feature Store.
• Amazon VPC / PrivateLink for secure ML endpoint networking.
• Data governance & compliance experience (PII / GDPR).
• Infrastructure as Code (Terraform / CloudFormation / CDK).