Job Description

HFM is an internationally acclaimed multi-asset broker, delivering cutting-edge trading tools, platforms, and conditions to traders worldwide. We are committed to innovation, transparency, and excellence in the financial markets.

As we continue to expand, we are seeking a driven and strategic Head of AI Engineering to join our team to lead our AI engineering team at HFM. This is a hands-on role where you will lead the engineers who build the company's AI capability, find where AI removes real effort across our technology and business functions, and turn those opportunities into systems that run in production, are measured against a baseline agreed before the work starts, and stay there. The role carries the AI Engineering Lead accountabilities set out in the Group AI Policy, including release approval for our highest-risk AI systems.

Your role at HFM

As Head of AI Engineering, you will report to the Chief AI Officer, with the Chief Technology Officer as functional owner. The engineering career ladder, architecture review and security review sit within IT, so you will work across both lines every week.

You will:

Lead a team of AI software engineers and AI process optimisation specialists: technical direction, code quality, craft-focused one to ones, and the development of each engineer.
Own the technical quality of what the team ships, and write the technical input into each engineer's performance review.
Find productivity opportunities: analyse workflows across IT and the business departments, and identify where AI removes manual effort at a scale worth building for.
Set the measurement before the build. Work with department heads to agree and sign a baseline for each process before the work begins, then report the result against it afterwards, including the forecasts that were missed.
Turn those opportunities into delivered systems, from design through to production support.
Hold the standards line, with evidence. Every system carries its risk classification, data classification, evaluation set, shadow-run result and audit trail, recorded in the tracking system at each gate. Company coding standards are followed, and technical debt goes down on a measure you can show.
Run the delivery queue. Requests arrive from every part of the business and land in a single prioritised backlog. You will score, size and sequence that work, and say clearly what is not being done.
Enforce permissions in the data layer, never by instructing a model. All model access runs through a single route, with access rights applied in the data and retrieval layer and full audit logging behind it.
Take designs through architecture and security review early, and bring the constraints back before the team builds against the wrong assumption.
Own production practice for your team's services: runbooks, monitoring, on-call participation and post-incident follow-through.
Own the internal AI platform: prompt libraries, evaluation harnesses, reusable agents and shared components, built so that each new automation costs less to deliver than the last.
Evaluate models, tools and vendors on evidence: benchmarks, cost per unit of work, failure modes and data handling. Know what each system costs to run once it is live, so that a system whose benefit no longer clears its running cost can be identified and retired.
Plan capacity and make the case for the team. Maintain a view of delivery capacity against forecast demand, and produce the annual hiring proposal that follows from it.
Change the method, not only the systems. We expect this role to re-examine how AI work is delivered here: how quickly new models are adopted, which agents and tools the team standardises on, and how the delivery framework itself should be restructured. We expect you to propose those changes rather than work around them.
Be the single technical point of contact for delivery and technical gates between the AI function and IT, and mentor engineers and colleagues across the company in using AI well.

Requirements

6+ years experience building software or data systems, including 2+ years leading engineers as a manager or technical lead.
Real AI depth demonstrated in production work: machine learning models, large language models or generative AI solving actual business problems, not prototypes alone.
Strong Python, and production experience with the large language model stack: building services and application programming interfaces, orchestrating agents and tool use, and working with the major model providers. We care that you have shipped these systems, not which particular libraries you used.
Production experience with retrieval augmented generation and vector search, built to be secure, efficient and cost aware.
Containerised environments (Docker), cloud deployment and CI/CD pipelines as everyday working material.
Evaluation discipline: how a model is tested, how regression is caught, and how latency, cost, accuracy, hallucination and model drift are measured and reported.
Practical command of retrieval, agents and tool use, and prompt engineering, with a clear view of when fine tuning is the right answer and when it is not.
DevOps and IT automation experience: integrating AI into CI/CD pipelines, infrastructure automation and workflow tooling, with REST APIs and AI-driven microservices.
Cloud AI and ML services, AWS preferred, and command of the cost model behind them: token and inference cost at pilot scale and at full adoption.
Data protection in a regulated environment: what may leave the company, what must stay inside, and what has to be evidenced afterwards.
Able to hold a quality bar with senior peers without becoming the bottleneck, and to explain an AI limitation or risk in plain terms to decision makers.
Experience evidencing benefit to a finance or business audience: agreeing a baseline before the work, measuring the result afterwards, and reporting it honestly when it falls short.
Comfortable working across a dual reporting line, where architecture and security approvals sit in another function.
Financial services or another regulated industry is an advantage. Trading domain knowledge is not required.
Experience training models with PyTorch or TensorFlow is an advantage, not a requirement. Our work is applied: integrating, retrieving, orchestrating and evaluating, rather than training models from scratch.

Resumes must be submitted in English.
Applicants must be eligible or have legal authorization to work in the country where the position is based.