This a curated list of projects I no longer work on that I am proud of and/or display my research and product interests. A less curated but more complete list can be found here: All past projects

2023

August ‘23 to January ‘24: Jubmoji

April ‘23 to July ‘23: ZuStamps

  • Funded by 0xPARC, led by me and collaborated with Rachel & Althea
  • Part of a long line of In-person cryptography projects I have done
  • First deployment of signature-producing NFC chips for ZK applications
  • Featured on Vitalik’s blog here

September ‘22 to February ‘23: WebAuthn and ZK research

  • Funded by Personae Labs, led by me and collaborated with Aayush and Richard
  • Culminated in heyauthn!
    • Demo using WebAuthn/Passkeys to power Semaphore identities
    • Later upstreamed to main Semaphore repo by Richard!
    • Used NFC CMAC chips to create Semaphore groups, first documented use of NFC for ZK
    • Posted to a ZK IAP class Discord channel
  • Supported work to create ZK circuits to prove knowledge of valid WebAuthn signature

2022

August ‘22 to January ‘23: Speeding up client-side zk-ECDSA

March ‘22 to November ‘22: heyanon!

  • Funded by 0xPARC and Personae Labs, collaborated with Lakshman Sankar
  • Anon twitter posting with ETH reputation
  • ECDSA proving was way too slow for this to be useful/fun, I was laser focused on improving that in later projects

2021 and before

September ‘20 to June ‘21: Adversarial learning & image model research

  • Researcher in Madry Lab, initially investigating adversarial learning as an ML application of my cryptography/security interests
  • Researched effect of “hidden augmentations” to learning models on ImageNet performance compared to actual architectural differences
  • Researched sample complexity of OOD performance for SoTA ImageNet models, as a follow-up to “Do Adversarially Robust ImageNet Models Transfer Better?”

December ‘18 to April ‘19: Blockchain quantitative research

  • Job at a (now defunct, lol) crypto hedge fund called Virgil Capital
  • Researched underlying cryptoeconomics of prominent DeFi platforms like MakerDAO and Compound to develop trading strategies

January ‘17 to August ‘18: Authenticated data structures research

  • Member of Devadas Lab at MIT, advised by Alin Tomescu
  • Researched and developed efficient append-only authenticated dictionaries, published in ACM CCS ‘19
  • Applications to certificate authorities and public-key directories
  • Clever combination of KZG polynomial commitments and binary tree structures