GPT-2 scale model implemented in pure C/CUDA from scratch.

A critical list of the most impactful vulnerabilities for Large Language Model applications.

GitHub organization containing the ce, ce-rs, rdev, and homebrew-ce tools.

A single-binary TUI for managing bare-metal servers without leaving the terminal.

A learning platform for decompiling GameCube games into matching C code.

A high-performance self-hosted photo and video management system.

A self-hosted real-time CRDT sync server supporting collaborative editing and AI agent integration.

A parametric, 100% offline 3D CAD modeler that runs in a single HTML file.

Every Guix package converted as a Nix flake.

An agentic memory framework using active associative reconstruction to reduce LLM token costs.

Implementation of the Autoregressive Boltzmann Generator framework for molecular sampling.

Code implementation for CAT-Q, a cost-efficient and accurate ternary quantization for LLMs.

A system for context recycling to extend LLM capabilities in long-horizon tasks without increasing context windows.

A fully transparent, ablation-driven post-training pipeline for 8B-scale LLMs.

An open-source project to replace the proprietary BIOS/UEFI firmware in most PCs.

A self-hostable, open-source alternative to Google Timeline

A collection of modular and reusable compiler and toolchain technologies.

A complete, cross-platform solution to record, convert and stream audio and video.

A modular, extensible and fast Kolmogorov-Arnold Network implementation.

Omnimodal world models for Physical AI, including code, checkpoints, and datasets.