I've been working in data science and ML for over 10 years, across financial services, marketing, and engineering. I get models out of notebooks and into production: training pipelines, scoring, deployment, monitoring, governance. More recently I've been integrating LLMs into live applications. I'm now looking for contract work.
I take ML projects from exploration through to something that's actually running in production, monitored, and meeting whatever governance bar the organisation needs. I've done this across financial services, marketing, and engineering. Here's roughly how it breaks down:
Training, validation, scoring, deployment, monitoring. I've built recommendation engines, anomaly detection systems, and personalisation platforms that run at scale. The kind of work where drift detection and retraining triggers matter because the model is actually touching customers.
I've built RAG systems, embedding pipelines, and prompt orchestration workflows, and wired Claude and GPT APIs into web applications with real users. Most of this has been in regulated settings, so evaluation, confidence scoring, and audit trails come as part of the package.
Deployment infrastructure, monitoring, data quality pipelines on AWS and GCP. I also deal with the governance side: audit logging, traceability, model risk docs. I've been in banking long enough to know what "production-ready" actually means when there's a compliance team looking over your shoulder.
Real systems, real users. Not slide decks that never went anywhere.
Recommendation system I built and productionised: automated training, scoring pipeline, deployment, Streamlit interface for the business team. Full governance and audit trail.
Built this end-to-end as technical founder. FastAPI backend, Claude API for campaign generation, dual scoring engine, GDPR-compliant unsubscribe system, Brevo engagement tracking.
Retrieval-augmented generation on Vertex AI and Cloud Run. Embedding pipelines, confidence scoring, evaluation logic. Designed to meet regulated environment requirements from day one.
Offer ranking and customer decisioning at scale. I co-led the ML side: ingestion through to low-latency inference at 100–150ms. Managed a team of 6.
Available for contract work, inside or outside IR35. Based in Birmingham. Remote preferred.