My Interests

I’m interested in cities, housing, transport, and the way technology shapes the built environment. A lot of my work has sat at the intersection of software, data, and the British planning system, so I’m especially drawn to questions about why we build so little, how institutions make decisions, and what better tools could do to improve the process.

I've been interested in technology, especially AI, from almost the moment I left school and was able to see work for what it is: the means to a prosperous end. We can, and should, automate the tasks we find arduous.

I like history, economics, and geography, especially when they help explain how places ended up the way they are.

My Career

I currently work at a prop tech start-up in London, where my team and I recently built and launched a platform for submitting planning applications. You can learn more about it here.

In 2023, I cofounded a start-up called Tract. We set out to build a land promoter capable of taking smaller parcels of land through the British planning system. We were able to raise a round of venture capital funding against this idea. We ended up pivoting towards building software for planning consultants. We ultimately decided that the market we were in would not deliver a return to our investors and returned their capital. You can read the full 8000-word postmortem and the lessons we learnt here

Before Tract, I worked as a data scientist and machine learning engineer at a few different places. I was one of the first users of the Transformers repository back when it was still called pytorch-pretrained-bert. I've spent a lot of time fine-tuning computer vision and language models for different tasks and working with messy tabular datasets.

I created Magpie (iOS, Android). It is an app for consuming podcasts, RSS feeds, newsletters, and articles saved from the web in one place. You can either read them or have them read to you via your on device voice, all properly connected to your phone's native media controls. You can also drive it via voice.