Skip to content
View RishiShah99's full-sized avatar

Highlights

  • Pro

Block or report RishiShah99

Block user

Prevent this user from interacting with your repositories and sending you notifications. Learn more about blocking users.

You must be logged in to block users.

Maximum 250 characters. Please don’t include any personal information such as legal names or email addresses. Markdown is supported. This note will only be visible to you.
Report abuse

Contact GitHub support about this user’s behavior. Learn more about reporting abuse.

Report abuse
RishiShah99/README.md

Hi there

About Me

  • I'm Rishi Shah, a Schulich Leader at McMaster University.

My Work

  • I've developed the supervised and reasoning layers of the current SOTA model for labour prediction at E3A Healthcare.
  • Engineered the first native Blackwell (tcgen05) training backward for the gated-linear-recurrence family
  • Developed a training method (RDR) that almost doubles a latent world model's forecast horizon

Reach out to me at rishishah994@gmail.com for any inquiries/collaborations.

Pinned Loading

  1. lethe lethe Public

    Native Blackwell (sm_100) tcgen05 training backward for the gated-linear-recurrence family (GDN-2/GLA/KDA/SSD), plus a contract-grade verifier that falsifies published GPU kernels. Six open Mamba-3…

    Python 10

  2. RDR RDR Public

    Reference implementation for the paper "Rollout-Decoded Reconstruction for Latent World Models": one loss term that trains a latent world model's decoder on its own free-running rollout, at zero ad…

    Python

  3. cgm-on-chip cgm-on-chip Public

    This repository reimplements OSDN and S4D in pure C++ with their own autograd engine, trained on real clinical data, and run on a $4 microcontroller to predict hypoglycemia 60 minutes ahead.

    C++ 1

  4. Sentinel Sentinel Public

    1.2K downloads. Hardware-aware diagnostics and intelligent memory analysis for embedded systems development. A VS Code extension that validates your code before compilation, catching hardware-speci…

    JavaScript