Senior MLOps Engineer — Hybrid London, Scalable AI

Harnham - Data and Analytics Recruitment; Greater London; Hybrid; GBP 51,000 - 85,000; Full time

Benefits offered by this job

Hybrid work model

Job summary

Harnham is seeking a Senior MLOps Engineer in London (Hybrid) to scale and stabilise our ML platforms. You will build and evolve cloud infrastructure, deploy ML models with Azure ML and AKS, and design robust pipelines for training, deployment and retraining.

Your work will emphasize reliability, security, and governance across client environments. You will collaborate with data scientists and engineers to implement IaC, monitoring, and observability using Prometheus, Grafana, and related tools,

Qualifications

• Strong experience in MLOps/Platform Engineering for ML systems.
• Advanced Python programming with a software engineering mindset.
• Hands-on experience with Azure-native MLOps services, AKS, and model deployment pipelines.

Responsibilities

• Build, maintain, and evolve cloud infrastructure for ML/AI services in production.
• Deploy, manage, and monitor ML models using Azure ML and AKS.
• Design orchestration pipelines for training, deployment, inference, and retraining workflows.
• Implement infrastructure-as-code with Terraform, Bicep, or Pulumi.

Skills

MLOps

Platform Engineering

Python

Azure MLOps

Kubernetes

CI/CD

Terraform/Bicep

Observability

Dagster/Airflow/Prefect

Azure DevOps

Tools

Azure ML

AKS (Azure Kubernetes Service)

Terraform

Bicep

Pulumi

Dagster

Airflow

Prefect

Prometheus

Grafana

Azure Monitor

Job description

Harnham is seeking a Senior MLOps Engineer in London (Hybrid) to scale and stabilise our ML platforms. You will build and evolve cloud infrastructure, deploy ML models with Azure ML and AKS, and design robust pipelines for training, deployment and retraining.

Your work will emphasize reliability, security, and governance across client environments. You will collaborate with data scientists and engineers to implement IaC, monitoring, and observability using Prometheus, Grafana, and related tools,