Machine Learning Engineer

Machine Learning Engineer Model Deployment \& Inference Performance

Established financial services institution

• Mid to senior level

Building a model is one thing. Making it fast, efficient and reliable in production is another.

Our client is running ML at serious scale across risk, payments and customer-facing products, and they’re looking for an engineer who wants to solve the difficult bit: getting models to perform when the real-world constraints kick in.

You’ll work with data scientists and engineering teams to take models from development into production, optimise how they run, and keep them performing once they’re live.

The challenge is interesting because you can’t simply throw more compute at the problem. A significant part of the estate handles single requests, tight latency targets and fixed compute constraints.

That means profiling the bottleneck, understanding what’s really happening under the hood, making the change and proving it worked.

What you’ll be doing

• Take ML models from development into production
• Work closely with data scientists, platform and engineering teams
• Optimise inference through profiling, quantisation, graph optimisation, threading, memory and runtime tuning
• Build and improve model, data and deployment pipelines
• Benchmark performance and measure the impact of your changes
• Monitor latency, availability, model performance and drift
• Help shape reliable, scalable and secure production services
• Support retraining and the wider model lifecycle
• Work within the governance and controls expected in financial services
• Contribute to CI/CD, testing and code reviews

What we’re looking for

• Strong experience deploying and maintaining ML models in production
• Genuine inference optimisation experience, backed by measurable results
• You can explain the baseline, where the bottleneck was, what you changed and what improved
• Experience with quantisation and accuracy recovery
• Knowledge of multiple inference runtimes
• Strong production Python
• PyTorch or TensorFlow, plus experience with gradient-boosting models
• Docker, Kubernetes and CI/CD
• Experience with AWS, Azure or similar cloud environments
• Good understanding of observability and production engineering

Nice to have

• C++, Rust or Java used in performance-critical production code
• Vectorisation, cache behaviour, memory bandwidth and thread/core affinity
• Multi-threaded or distributed systems
• Real-time or streaming inference
• Experience working with fixed compute and demanding latency targets
• Kernel-level optimisation or systems programming
• Financial services or another regulated environment

For senior candidates, we want to hear about something you’ve actually made faster.

What was the problem? What did you change? And how much faster did it get?

What this isn’t

This isn’t a research role or another generic MLOps position.

It’s for someone who enjoys getting into the detail, finding the bottleneck and making production ML faster, leaner and better.

If that sounds like your kind of engineering, we should talk.

justin.toomey@datatech.org.uk