ML Ops​/Data Engineer

Position: ML Ops / Data Engineer
Location: Greater London

ML Ops / Data Engineer page is loaded## ML Ops / Data Engineer locations:
London time type:
Full time posted on:
Posted Yesterday job requisition :
CMC
5104#
** ML Ops / Data Engineer
**** Role Overview
** We’re hiring an
** ML Ops Engineer / Data Engineer
** to own the reliability, scalability, and operational integrity of our machine-learning systems in research & production. This role sits at the intersection of data engineering and ML infrastructure: you’ll design and operate data pipelines that feed models, and you’ll build the tooling that trains, deploys, monitors, and retrains them.

You’ll work closely with research engineers and product teams, taking models from experimentation to production-grade systems with clear SLAs, reproducibility guarantees, and observable behaviour. This is not a research role; it is a hands-on engineering role focused on making ML systems work reliably at scale.
** What You’ll Work On
**** ML lifecycle infrastructure
*** Product ionizing models: packaging, deployment, versioning, and rollback
* Designing CI/CD pipelines for ML (training → validation → deployment)
* Implementing model monitoring (data drift, prediction drift, performance decay)
* Managing experiment tracking and reproducibility
** Data engineering foundations
*** Building and maintaining batch and near–real-time data pipelines
* Ensuring data quality, schema evolution, and lineage across systems
* Designing datasets and feature pipelines that support both training and inference
* Operating pipelines with clear reliability and latency expectations
** Operational ownership
*** Defining and meeting availability, latency, and freshness targets for ML services
* Debugging production issues across data, infrastructure, and model layers
* Improving system robustness through automation and observability
* Collaborating with platform and security teams on access, secrets, and compliance
** Engineering rigor
*** Writing production-grade Python used in long-running services and pipelines
* Establishing testing, validation, and release practices for ML systems
* Making trade-offs explicit between research flexibility and production stability
** Required Qualifications
*** 3–7 years of professional experience in
** ML Ops, Data Engineering, or adjacent backend roles
*** Strong
** production Python
** skills (clean APIs, testing, performance awareness)
* Experience deploying and operating ML models in production environments
** Solid understanding of:
*** Model training vs. inference requirements
* Batch vs. streaming data pipelines
* Failure modes in data-driven systems
* Hands-on experience with at least one modern orchestration or workflow system
* Comfort working with cloud infrastructure and containerized workloads
* Ability to reason about system design, not just tool usage
** Nice-to-Have
*** Experience operating systems at TB-scale data volumes or higher
* Prior ownership of model monitoring, drift detection, or automated retraining
* Familiarity with feature stores or online/offline feature consistency problems
* Experience supporting multiple models or teams on a shared ML platform
* Exposure to regulated or high-reliability production environments
** Tech Stack (Current & Expected Evolution)
***
* Languages:

** Python (core)
** ML & Data:
** PyTorch / similar frameworks, experiment tracking, structured datasets
** Pipelines & Orchestration:
** Workflow schedulers for batch and near-real-time processing
** Deployment:
** Containers, model serving frameworks, infrastructure-as-code
** Observability:
** Metrics, logging, and alerting across data and model layers
** Cloud:
** Managed compute, storage, and networking (provider-agnostic mindset)
The stack will evolve. We value engineers who understand
** why
* * systems are built a certain way and can adapt tools as requirements change.
** Why This Role Matters
** Our models only create value when they are
** correct, observable, and dependable in production.
** This role is responsible for that reality. You’ll reduce the gap between promising experiments and systems that can be trusted by downstream products and customers.

If you care about data correctness, operational clarity, and building ML systems that don’t…