Role purpose

Build and operate the shared infrastructure that enables AI and ML systems to move reliably from development into production. Provide reproducible deployment, model and configuration versioning, monitoring, security and cost visibility across predictive models and generative AI services. Ensure applications remain observable, recoverable and supportable during brokerage operating hours.

Key responsibilities

Platform foundations and deployment

• Design and maintain development, testing and production environments for model training, batch scoring, online inference and LLM-enabled applications.
• Build CI/CD pipelines with artifact versioning, automated checks, environment promotion, approvals and tested rollback for code, models, prompts and configuration.
• Implement model registries, experiment tracking and reproducible execution environments, linking deployed artifacts to their code, data and evaluation records.
• Operate model-serving endpoints, approved model gateways and container workloads; enable controlled access to third-party and internally hosted models.

Observability and reliability

• Monitor latency, throughput, availability, error rates, resource utilisation and model or token costs across AI services.
• Provide instrumentation for data drift and model quality signals in collaboration with model owners, separating infrastructure alerts from analytical performance alerts.
• Implement autoscaling, quotas, rate limits, timeouts, fallback behaviour and graceful degradation appropriate to each service’s operating requirements.
• Define service objectives, runbooks and recovery procedures; test backups, restoration and deployment rollback and participate in the agreed incident response rota.

Security and operational efficiency

• Apply infrastructure as code, least-privilege access, secrets management, network controls and approved data residency settings with Security and Infrastructure teams.
• Protect training data, model artifacts and logs through access restrictions, encryption and retention controls; prevent secrets or unnecessary client data from entering telemetry.
• Optimise CPU/GPU capacity, inference configurations and cloud spend, making application and model costs visible to technical and business owners.
• Review provider changes, dependency vulnerabilities and platform lifecycle risks; maintain platform documentation and reusable deployment patterns for engineering teams.

Essential experience and qualifications

• 5+ years in DevOps, SRE, platform engineering or ML infrastructure, including deployment and operation of AI or ML workloads.
• Strong experience with a major cloud platform, containers, infrastructure as code and CI/CD.
• Python or comparable scripting skills and practical Linux, networking, identity and secrets-management knowledge.
• Experience with model serving, experiment tracking, artifact registries and reproducible deployment practices.
• Experience building observability, capacity planning, incident response and recovery processes for production services.
• Understanding of the operating characteristics of LLM APIs and model inference, including variable latency, usage limits and cost controls.
• Bachelor’s degree in Computer Science, Engineering or a related field, or equivalent infrastructure engineering experience.

Preferred experience

• Experience with Kubernetes, MLflow, model-serving frameworks and GPU workload optimisation where relevant to the chosen architecture.
• Experience with financial services infrastructure, 24/5 trading environments and support for continuously available client systems.
• Experience with private AI deployments, model gateways and multi-provider resilience.