Simon Eskildsen, co-founder and CEO of turbopuffer, built his engineering career on a single discipline: napkin math. The practice, quick back-of-envelope calculations against theoretical hardware limits, exposed that existing vector search solutions were charging far more than the underlying compute costs justified. That gap became turbopuffer. The company landed Cursor as its first customer after raising an $8M seed round, with Cursor betting on the team before the product was fully built.
The original interview, recorded at the AI Engineer's World Fair and running 55 minutes, traces how Eskildsen went from competing in the International Olympiad for Informatics as a Danish high schooler to spending nearly a decade on infrastructure at Shopify. The IOI section alone is worth reading: Eskildsen entered without knowing what NP-complete meant, survived a brutal weekend bootcamp covering divide-and-conquer and dynamic programming on a four-hour rotation, and still claimed one of six spots on the Danish national team. That pattern, finding the ceiling of what he knew and then pushing through it, repeats at every stage of his career.
The piece also breaks down Eskildsen's taxonomy of why founders actually raise venture capital: funding R&D, funding growth, and stroking egos. He is direct that the third reason drives more rounds than anyone admits. If you work on infrastructure, build developer tools, or are trying to understand how a founder thinks about cost and compute limits at the systems level, read this in full.
[READ ORIGINAL →]