Skip to content
@SaturdayLabs

Saturday Labs

╔═══════════════════════════════════════════════════════════════════════╗
║                                                                       ║
║  ███████╗ █████╗ ████████╗██╗   ██╗██████╗ ██████╗  █████╗ ██╗   ██╗  ║
║  ██╔════╝██╔══██╗╚══██╔══╝██║   ██║██╔══██╗██╔══██╗██╔══██╗╚██╗ ██╔╝  ║
║  ███████╗███████║   ██║   ██║   ██║██████╔╝██║  ██║███████║ ╚████╔╝   ║
║  ╚════██║██╔══██║   ██║   ██║   ██║██╔══██╗██║  ██║██╔══██║  ╚██╔╝    ║
║  ███████║██║  ██║   ██║   ╚██████╔╝██║  ██║██████╔╝██║  ██║   ██║     ║
║  ╚══════╝╚═╝  ╚═╝   ╚═╝    ╚═════╝ ╚═╝  ╚═╝╚═════╝ ╚═╝  ╚═╝   ╚═╝     ║
║                                                                       ║
║                   ██╗      █████╗ ██████╗ ███████╗                    ║
║                   ██║     ██╔══██╗██╔══██╗██╔════╝                    ║
║                   ██║     ███████║██████╔╝███████╗                    ║
║                   ██║     ██╔══██║██╔══██╗╚════██║                    ║
║                   ███████╗██║  ██║██████╔╝███████║                    ║
║                   ╚══════╝╚═╝  ╚═╝╚═════╝ ╚══════╝                    ║
║                                                                       ║
║                      proofs · fields · kernels.                       ║
║                                                                       ║
╚═══════════════════════════════════════════════════════════════════════╝

> 64K FIELD ELEMENTS FREE

Cryptography, proof systems and high-performance kernels, built from first principles.



Saturday Labs is an open research and engineering lab. We pick a primitive, a protocol or a kernel, derive it on paper, and implement it from scratch. Every implementation is tested against a reference and benchmarked. We publish the code and what we learned from building it.

Focus areas

  • Zero-knowledge proof systems: sumcheck, GKR, FRI, WHIR, PLONK, folding schemes
  • Field and polynomial arithmetic: finite fields, extension fields, NTTs, multilinear extensions
  • GPU computing: parallel primitives, matrix multiplication, fused kernels for ML and cryptography
  • Where they meet: GPU-accelerated provers, and verifiable computation for machine learning

How we work

  • First principles. Understand why something works before optimizing it.
  • Reference first. A naive, obviously correct version comes before the fast one.
  • Measured. If it's described as fast, there's a benchmark to back it up.
  • In the open. Code, notes and mistakes are all public.



Built on Saturdays.

Pinned Loading

  1. how-to-plonk-solved how-to-plonk-solved Public

    Fully solved solutions to how to plonk (created by zk security team)

    Jupyter Notebook

Repositories

Showing 7 of 7 repositories

Top languages

Loading…

Most used topics

Loading…