I build machine learning systems that help organizations better understand customers, improve retention, increase customer lifetime value, and make better decisions.
I am a Machine Learning Engineer at Spotify, currently based in London.
My experience spans personalization, customer understanding, experimentation, retention, and growth. I've worked on production machine learning systems and data products that influence customer experiences and business outcomes at scale.
I studied Statistics at the University of Warwick and have a strong interest in causal machine learning, constrained optimization, and decision-making under uncertainty.
Outside of work, I enjoy exploring how experimentation, machine learning, and decision systems can drive better outcomes.
Spotify
Building machine learning systems to personalize subscription grace periods, improving retention and net revenue.
Trainline
Worked on customer lifetime value modelling and profitable growth initiatives, including predictive modelling and contextual bandits for conversion optimization.
Guidehouse
Digital Twins for Gas Distribution Network.
Dept. of Medicine, University of Hong Kong
Image Classification for Orthopaedics.
Tailoring product experiences to individual customers using behavioural signals and machine learning.
Models and pipelines that support acquisition, conversion, and lifetime value across the customer journey.
Designing and analysing A/B tests so product and business teams can make confident decisions.
Connecting predictions to actions — turning model outputs into policies, thresholds, and business rules.
Estimating real effects of interventions when randomised experiments aren't feasible or affordable.
Relational foundation models and other enterprise FMs that predict directly from business data — and how to extend them from zero-shot predictions to zero-shot actions.
Thoughts on machine learning, experimentation, growth, and decision-making.
I benchmarked zero-shot and continued-pretrained Relational Transformers against XGBoost and RelGT on two RelBench tasks — 18 hours on a 48GB Mac. One task was competitive; the other collapsed.
Relational foundation models learn directly from the relational data businesses already have — no hand-built feature pipelines. Here's why that shifts the ML stack.
TabFMs and Kumo's RFMs are quietly automating feature engineering and model training. What's left for MLEs? Policy and decision-making.