MLOps Engineer

MLOps Engineer
Job Description
Field
Details
Chapter area
AI (AI Engineering chapter)
Reports to
Chapter Lead – AI Engineering
Works with
AI Engineers, Data Engineers, AI Architects, AI QA, AI Model Risk SME, Run / Managed Services team
Experience
3–6 years in MLOps, DevOps or ML engineering, with production deployment experience
About the Role
The MLOps Engineer builds and operates the pipelines, infrastructure and tooling that take ML and GenAI solutions from development into reliable, governed production. The role owns deployment automation, monitoring and release discipline across the build-to-run lifecycle, and bridges AI Engineering, Data Engineering and platform/DevOps. The role covers both classical ML and LLMOps: evaluation, prompt and model versioning, observability and cost control for LLM-based and agentic systems.
Key Responsibilities
Deployment and release automation
Build and maintain CI/CD and continuous-training pipelines for ML and GenAI components, with automated testing, approval gates, safe rollout (canary / blue-green) and tested rollback paths.
Containerise and deploy model, RAG and agent endpoints on Kubernetes-based infrastructure, with environments provisioned as code.
Manage model and prompt registries, versioning, experiment tracking and lineage so every release is reproducible and traceable.
Monitoring, observability and reliability
Implement monitoring, logging and alerting for latency, availability, drift, output quality (groundedness, hallucination) and safety, with agreed thresholds and dashboards.
Define runbooks, support incident response and root-cause analysis, and lead build-to-run handover and hypercare with the Run team.
LLMOps and GenAI enablement
Operationalise RAG pipelines, vector stores, prompt libraries and automated evaluation suites; enforce "no eval suite and rollback path, no release".
Track and optimise token usage, model routing and compute cost, with tagging that supports usage reporting and chargeback.
Implement runtime guardrails (for example PII masking and prompt-injection defence) in partnership with Security and the AI Architects.
Governance, security and Responsible AI
Build AI delivery lifecycle gate evidence (test results, approvals, model and prompt versions, audit logs) into pipelines so it is captured by default.
Apply least-privilege access, secrets management, data-residency and no-production-data-in-development controls.
Implement monitoring thresholds and Responsible AI checks agreed with the AI Model Risk SME.
Collaboration and reuse
Work with AI Engineers and Data Engineers to productionise models and GenAI applications reliably.
Contribute reusable pipeline templates, patterns and documentation to the AI CoE reuse library; coach squad members on MLOps practice.
Key Skills & Experience
Qualification
Details
Education
Bachelor's degree in Computer Science, Engineering, Data Science or a related technical field
Core engineering (required)
Python and SQL; Docker and Kubernetes; CI/CD tooling (for example GitHub Actions, Azure DevOps or GitLab); infrastructure as code (Terraform)
MLOps platforms (required)
Model tracking, registry and orchestration using MLflow, Kubeflow, Airflow or cloud-native equivalents (Vertex AI, Azure ML); working knowledge of one ML framework (PyTorch, TensorFlow or scikit-learn)
Cloud (required)
Hands-on with Google Cloud (Vertex AI) or Azure, and working knowledge of the other
Monitoring (required)
ML and application observability: drift detection, performance dashboards, alerting (for example Prometheus / Grafana or cloud-native tooling)
LLMOps (required)
Working knowledge of RAG pipelines, vector databases, prompt versioning, LLM evaluation and tracing, and token / cost tracking
Governance awareness
Understanding of Responsible AI, model risk and audit-evidence expectations in a regulated environment
Collaboration
Works across AI, data, platform, security and risk teams; communicates clearly with technical and non-technical stakeholders
Preferred Experience
End-to-end ML or GenAI deployment pipelines in a regulated industry (insurance, banking or financial services).
GPU resource management and inference optimisation for LLM serving.
Agent observability and tracing; feature stores; model-serving frameworks.
Experience with ITSM incident and change processes (for example ServiceNow) applied to AI services.
Certifications (a plus, not required): Google Professional Machine Learning Engineer, Microsoft Azure AI certification (for example AI-300 or Azure AI Engineer Associate), Certified Kubernetes Administrator.

Work arrangement
No