Notes
Notes for quant and AI interviews
A 315-page reference I wrote for myself while I was interviewing in 2026. Probability through market microstructure, every result proved, 191 problems with full solutions. It's free. If you're trying to break into these roles it's the thing I'd have wanted at the start.
What's in it
Eight parts: probability and games of chance; statistics and regression; linear algebra and calculus; machine learning and deep learning; trading, portfolios, market making and market structure; interview craft; computer science for research engineers; then the problem bank and its solutions.
It's built like a textbook rather than a cheat sheet. There are 143 proofs and 145 worked examples. Alongside them, three running commentaries say why a result is true without symbols, what an interviewer is testing with it, and where it shows up in a real book.
Where the questions came from
A star marks a question reproduced from a live 2026 interview, a diamond one from a timed online assessment. There are 57 in the problem bank and thirteen more in the body text.
No firm is named anywhere in the document for obvious reasons. For the record, and kept out of the document itself, the firms I'd sat an assessment or an interview with by the September revision were Akuna Capital, Maven Securities, Qube Research & Technologies, Susquehanna, Wintermute, Anthropic, Mistral AI and Orbital Industries. The questions are representative of what gets asked for quant and research roles at firms like these.
Caveats
The coverage follows the roles I applied for, so market making and crypto market structure are heavier than they would be in a balanced treatment, and there's almost nothing on stochastic calculus or C++. The proofs are mine and the versions are explained in the way that made them click for me, so if a step looks wrong it might be. Tell me and I'll fix it.