0xPARC

I spent one year working at 0xPARC, a research organization dedicated to advancing programmable cryptography. Here are a few artifacts from my time there.

FHE Guide

I wrote an extensive guide to FHE for founders, engineers, researchers, and anyone interested in cryptography to better understand fully homomorphic encryption at both a high level and a technical level.

Take-home challenge

I wrote a coding and math challenge for 0xPARC as part of the interview process. The task is to optimize polynomial evaluation under a cost model inspired by CKKS: additions are free, multiplications cost a level, and you only have so many levels to spend.

It's deliberately open-ended — there's no single correct answer, and the more time you put in, the more you can optimize. Feel free to clone it and try it yourself! If you believe you have substantial progress, please email me at hello@holdenmui.com!

Polynomial computer talk

In Fall 2025, I gave a talk on building an encrypted computer out of polynomials at the 0xPARC Future of (Super) Computing event, jointly hosted with the MIT Informatics Tournament.

PLoNKathon

For the same 0xPARC Future of (Super) Computing event, I designed a puzzle event called PLoNKathon. Teams of three or four have 90 minutes to crack five puzzles, but solving them is only half the game. You also score points by getting other teams to sign attestation slips certifying that you solved a problem... whether or not you actually did.

The twist is that bluffing is explicitly allowed. If you can't solve a puzzle, your best move is to trick another team into believing you did, executing a Plausible Lie of Non-Knowledge (PLoNK). The whole event is a social analogue of a zero-knowledge proof system: you're proving that you know a solution without revealing it... except here, the verifiers are humans, and dishonest provers are part of the game.