Jean-Luc Watson

PhD student at UC Berkeley
jlw@berkeley.edu |

I am a graduating computer science PhD student at UC Berkeley, co-advised by Raluca Ada Popa and Prabal Dutta. I am a recipient of the National Defense Science & Engineering Graduate (NDSEG) Fellowship (2018), and I completed my B.S. and M.S. in Computer Science at Stanford University, advised by Phil Levis.

My PhD has centered on building systems for privacy and security on constrained devices. I focus on using GPUs to practically accelerate cryptographic protocols like multi-party computation, supporting secure machine learning. I've also worked on embedded systems, enhancing mobile sensor networks with strong privacy guarantees.

Recent News

[thesis]

[slides]

Projects

Piranha

Accelerate secure machine learning.

An open-source platform for multi-party computation (MPC) protocol development supporting GPU acceleration, supporting three linear secret-sharing schemes. We used Piranha to securely train a large neural network in less than a day, 10x faster than prior work.

Publications

Piranha: A GPU Platform for Secure Computation

USENIX Security Symposium 2022 (USENIX Security '22)

Code

ucbrise/piranha

C++ implementation of GPU-enabled MPC protocols and NN training.

Slides

USENIX Security '22

August 2022

Nebula

Privately sense the world around us.

A privacy-preserving protocol to retrieve data from sensors deployed almost anywhere, using third-party mobile phones. We developed a prototype using DTLS-over-BLE on Nordic microprocessors to demonstrate security and low-power capability.

Publications

Nebula: A Privacy-First Platform for Data Backhaul

IEEE Symposium on Security and Privacy 2024 (Oakland '24)

Where the Sidewalk Ends: Privacy of Opportunistic Backhaul

European Workshop on Systems Security 2022 (EuroSec '22)

Code

lab11/nebula

Nebula protocol implementation.

Slides

Oakland '24

May 2024

EuroSec '22

April 2022

Accelerating Zero-Knowledge

A GPU-based approach to high-throughput proving.

Pipeline for high-throughput zero-knowledge proving using GPUs, leveraging off-device memory to support more and larger proofs than prior techniques.

Certified Differential Privacy

Ensure databases correctly apply differential privacy.

Prototype of a certifiable differential privacy protocol in Rust where a data provider can prove that differential noise was correctly applied to a query. We show practical performance, with queries on a Census dataset verified in 40μs.

Publications

Certifying Private Probabilistic Mechanisms

44th Annual International Cryptology Conference (CRYPTO '24) (to appear)

Code

jlwatson/certified-dp

A Rust-based cryptographic protocol to verify differentially-private queries.

Retcon

Patch embedded devices in a heartbeat.

A novel compile-time static analysis pipeline for event-driven C programs that can update embedded devices “live” in less than a millisecond -- without missing a beat -- while avoiding state loss.

Publications

Retcon: Live Updates for Embedded Event-Driven Applications

International Conference on Information Processing in Sensor Networks 2024 (IPSN '24)

Code

lab11/live-update

Live update pipeline implementation using Zephyr RTOS for the nRF9160.

Slides

IPSN '24

May 2024

Embedded Operating Systems

New models for embedded computation.

Specified and formalized the Tock embedded operating system's multi-tiered security model, as well as made a case for the next generation of embedded OSes that will have to be distributed amongst many chips.

Publications

Tiered Trust for Useful Embedded Systems Security

European Workshop on Systems Security 2022 (EuroSec '22)

Embedded OSes Must Embrace Distributed Computing

International Workshop on Next-Generation Operating Systems for Cyber-Physical Systems 2019 (NGOSCPS '19)

Publications

Certifying Private Probabilistic Mechanisms [paper] [code]
Zoe Ruha Bell, Shafi Goldwasser, Michael P. Kim, Jean-Luc Watson
44th International Cryptology Conference (CRYPTO 24)

Retcon: Live Updates for Embedded Event-Driven Applications [paper] [code] [slides]
Jean-Luc Watson, Saharsh Agrawal, Ryan Tsang, Sherry Luo, Raluca Ada Popa, and Prabal Dutta
ACM/IEEE Intl. Conference on Information Processing in Sensor Networks (IPSN) 2024

Nebula: A Privacy-First Platform for Data Backhaul [paper] [code] [slides]
Jean-Luc Watson, Tess Despres, Alvin Tan, Shishir G. Patil, Prabal Dutta, and Raluca Ada Popa
IEEE Symposium on Security and Privacy 2024

Piranha: A GPU Platform for Secure Computation [paper] [code] [slides]
Jean-Luc Watson, Sameer Wagh, and Raluca Ada Popa
USENIX Security '22

Where the Sidewalk Ends: Privacy of Opportunistic Backhaul [paper] [slides]
Tess Despres, Shishir Patil, Alvin Tan, Jean-Luc Watson, and Prabal Dutta
EuroSec '22

Tiered Trust for Useful Embedded Systems Security [paper]
Hudson Ayers, Prabal Dutta, Philip Levis, Amit Levy, Pat Pannuto, Johnathan Van Why, and Jean-Luc Watson
EuroSec '22

Embedded OSes Must Embrace Distributed Computing [paper]
Branden Ghena, Jean-Luc Watson, and Prabal Dutta
NGOSCPS '19

Teaching

EE375: Teaching Techniques for Electrical Engineering
Graduate Student Instructor (Fall 2021, Fall 2023, Spring 2024)

EECS149/249A Introduction to Embedded Systems
Graduate Student Instructor (Fall 2019)

CS107E Computer Systems from the Ground Up
Course Assistant (Spring 2018)

CS198B Additional Topics in Teaching Computer Science
Co-Instructor (Winter 2017, Spring 2017, Autumn 2017, Winter 2018)

CS198 Teaching Computer Science
Co-Instructor (Winter 2017, Spring 2017, Autumn 2017, Winter 2018)

CS106X Programming Abstractions in C++
Section Leader (Autumn 2016)

CS106B Programming Abstractions
Section Leader (Spring 2015, Winter 2016, Spring 2016)

CS106A Programming Methodologies
Section Leader (Winter 2015, Autumn 2015)