I'm a student working toward research in physics-informed machine learning — the area where numerical methods for differential equations meet neural networks.

What drew me to it is that it sits on top of real mathematics rather than replacing it. A physics-informed network doesn't learn a solution from data alone; it's constrained by the equations that govern the system. That makes the maths worth understanding properly rather than treating as background.

How I'm going about it#

I'm learning from first principles and in order — mathematical foundations, then machine learning fundamentals, then implementing things from scratch rather than assembling them from libraries. It's slower than following tutorials, and the point is that it's slower: I'd rather be able to derive something than recognise it.

That work is public, unfinished, and organised by phase. It's currently at the beginning.

I'm also reading my way through quantitative finance independently. It's an interest rather than something I've built in yet — the appeal is similar, in that the useful parts are the ones with real structure underneath.

What's on this site#

• Now — what I'm actually studying and reading this week. The most current thing here.
• Projects — things I've built, added as they become real rather than in advance.
• Research — notes and writeups. Empty for now; it'll fill as I have something worth saying.

I'd rather this page be accurate than impressive. If something here interests you, or you think I've got something wrong, the links in the footer work.