TxsharDev

Follow

🎯

Focusing

Tushar Sharma TxsharDev

🎯

Focusing

Follow

I write code at every level of the stack. Even the ones that void your warranty.

3 followers · 5 following

United States
06:19 (UTC -05:00)
@TxsharDev

TxsharDev/README.md

Tushar Sharma

Full-stack from the browser to the bare metal.

I build web apps, systems tools, CUDA kernels, and AI inference engines. If there's a layer, I've written code in it.

The Stack

Web & Cloud · Production apps, APIs, infrastructure.
Systems · Rust binaries that launch before your finger leaves the key.
GPU / CUDA · Custom kernels, hand-tuned PTX, where a single cache miss costs you everything.
AI · Not using models. Building the engines that run them.

Currently

Something big. Stay tuned.

Tech

TypeScript React Next.js Node C# Python Rust C++ CUDA CUTLASS PTX

You don't learn the future by consuming it. You build it.

Pinned Loading

Foundry Foundry Public

Cast web into native metal. HTML/CSS/JS compiled to GPU-rendered native binaries.

Rust 1
Molten Molten Public

Write the math. Get the kernel. Fused CUDA kernel generation from mathematical specifications.

Python
Carbon Carbon Public

Same seed. Different GPU. Identical weights. Bit-exact deterministic training across GPU architectures.

Python
Redline Redline Public

Cross the performance line. Won't compile. Compile-time performance guarantees for Rust.

Rust
Sealed Sealed Public

Cross the supply chain. Cryptographic attestation, behavioral sandboxing, and trust verification for package installs.

Python
Pane Pane Public

GPU-aware system monitor for Windows. Per-process VRAM, dual GPU support, hardware controls. Single binary, Rust.

Rust