In this episode, Yitzy is joined by Yehuda Lindell—founder of Unbound Security (acquired by Coinbase), longtime professor and one of the earliest researchers in secure multiparty computation (MPC).
We trace his journey from Australia to Israel and from theory to industry, then dive into how Coinbase secures customer assets using layered cryptography: MPC, secure enclaves, HSMs, cold elements, and rigorous review processes. Yehuda explains MPC in plain English (with a great DNA example), why “defense-in-depth” beats any single fortress, how insider threats are modeled, and what Coinbase’s production bar looks like.
We also talk CS education in the age of AI, and why students should learn to program with AI rather than fear it. Yehuda closes with a pointer to CB-MPC, Coinbase’s open-source MPC engine, and a broader call for rigor when turning cryptography papers into production systems.
Resources and Links:
Yehuda on LI: https://www.linkedin.com/in/yehudalindell/
Yehuda on X: https://x.com/LindellYehuda
Coinbase's CB-MPC library: https://github.com/coinbase/cb-mpc
Timestamps:
00:00 — Intro & Yehuda’s background
05:43 — What is cryptography (for non-experts)
16:00 — MPC 101 (privacy, correctness, shard thresholds)
19:36 — Real-world key splitting; surviving full IT compromise
23:38 — Coinbase’s defense-in-depth philosophy
24:34 — What secure enclaves are (and side-channel realities)
28:52 — Cold elements, cross-domain guards, layered participants
31:31 — Coinbase’s production bar & independent reviews
35:54 — CS, AI, and training “with” AI
41:59 — CB-MPC and parting advice