I build production AI for the industrial world.

I'm a full-stack AI engineer: a decade of production AI for the chemical, manufacturing, software and automation sectors, built on an educational foundation in chemical engineering (bachelor's and master's) and data science (diploma). Models, infrastructure, apps and agents: the whole path from raw signal to a decision someone trusts. Currently at Navigance in Munich, where models I own optimise chemical plants around the world and alert plant operators to failures long before they occur.

Specific energy · manual control

A live simulation of what I do for a living. Try it, and watch for the occasional fault.

$50M+of value from predictive models running in Unilever factories worldwide
−3%energy on live ammonia plants, from optimisers I own today
20×R&D throughput, via the largest automation programme in Unilever's history
−90%design-time-to-insight on a $100M-a-year value stream at Koch

What I do

Modelling

From neural networks and Gaussian processes to gradient boosting, time series, NLP and first-principles models: whichever fits the problem, on noisy real-world data where a wrong prediction has a physical cost.

Software & infrastructure

Typed, tested, reviewed codebases and everything around them: MLOps, DevOps, CI/CD, deployment, monitoring, evaluation, retraining and health alerting. I lead engineering standards as well as write to them.

Apps & interfaces

Full-stack development of the optimisers, dashboards and calculators that operators and engineers use every day inside a commercial dashboard product, taken from notebook analysis to CI-validated production.

LLMs & agents

Retrieval-grounded assistants that answer from live process data, and agentic pipelines that carry a bug report through to a reviewed pull request. Practical, evaluated, and in production.

Experience

2024 — now
Data Scientist / ML Engineer — Navigance, Munich
End-to-end AI product ownership at a 22-person SaaS company whose dashboard product delivers advanced optimisation and detection systems to chemical plants worldwide. I develop across modelling techniques, including hybrid neural-network and first-principles reactor models for methanol, ammonia, maleic anhydride and syngas plants, extracting up to 3% in energy savings on live processes. I architect and develop the MLOps, DevOps and CI/CD stack (Airflow, MLflow, InfluxDB, PostgreSQL, Docker and more), consult international clients on AI initiatives, and build the LLM tooling, such as a process-expert chatbot grounded in live plant data, alongside the predictive health-alerting systems.
2021 — 2024
Engineer, Data Science & DevOps — Roke, Manchester
At a professional software house, on highly regulated government programmes, I delivered cloud-hosted ML products end to end: the models themselves (NLP, clustering, Gaussian processes) that let analysts work through billions of data points in minutes, and just as much of the software around them, from APIs, front ends and ETL to the Kubernetes infrastructure and CI/CD they ran on. Led the project team as SCRUM Master; Best Roke Project Team 2021 and 2022.
2021
Consultant Chemical Engineer — Koch Technology Solutions, Wilton
Developed the company-wide digitisation strategy for a $100M+/year technology value stream, and built the full-stack process automation behind it (Python, R, AWS, Docker, Aspen and AVEVA), cutting design-time-to-insight by 90%. Presented the strategy across the business and tendered contracts with third-party digitisation partners.
2018 — 2021
Senior Research Data Scientist — Unilever R&D, Liverpool
Predictive ML across global factories generating $50M+ of value, delivered full stack: from data sourcing and measurement techniques through to models behind cloud-deployed interactive GUIs. Delivered the largest R&D automation programme in Unilever's history, robotic platforms that raised throughput 20× and scaled data collection from hundreds to millions of points per day. Mentored students and led external university partnerships for robotics and data-science projects.
2017
Automation Engineer — Oman LNG (via SECL UK), Sur
Started as a placement student during my Master's and stayed on as Project Engineer: tendered, won and delivered ~$500K of control-systems projects on time at a live LNG plant, leading field teams of technicians, working alongside GE and Siemens contractors, and training Omani graduate engineers.

Before all of that: process design engineer, chemical scientist, elected student head of faculty, festival door supervisor, kitchen porter. Every one of them taught something.

Education & credentials

Education

Data Analyst Fellowship, BCS — Distinction, 2019–2020
MSc Process Modelling, University of Birmingham — Merit, 2017
BEng Process Engineering, Lancaster University — 2:1, 2016
Associate Member IChemE · RITTech

Languages & eligibility

English — native · German — B1/B2
UK citizen · EU Blue Card (Germany)

Toolbox

Typed, tested, production-ready. I pick the right tool for the job, and the jobs vary a lot.

Modelling: Neural networks · Gaussian processes · gradient boosting · first-principles and hybrid models · time series · NLP · clustering (TensorFlow, PyTorch, GPyTorch, scikit-learn)
AI systems: LLMs · RAG · multi-agent systems · MCP · evaluation harnesses
Platform & data: AWS · Airflow · Docker · Kubernetes · InfluxDB · PostgreSQL · Neo4j · Datadog · Linux · CI/CD · DVC
Languages & tools: Python · SQL · R · MATLAB · C++ · Dash · React · Aspen and AVEVA · UE4, and more when the job needs it

Beyond work

Hundreds of personal projects across data science, electronics, server management and agentic platforms; the experiments on this site are the presentable tip. Away from a keyboard: Lindy Hop and swing dancing, bouldering, and outdoor trad climbing. Formerly captain of an American football team.

Contact

Open to senior roles across industrial AI: applied science, ML engineering, agentic systems, and technical leadership.
Reach me on LinkedIn.