Skip to content
View maci0's full-sized avatar
46 followers · 37 following
  • Red Hat
  • Singapore

Achievements

Achievement: YOLOAchievement: QuickdrawAchievement: Arctic Code Vault ContributorAchievement: Pull Sharkx2Achievement: Starstruck

Achievements

Achievement: YOLOAchievement: QuickdrawAchievement: Arctic Code Vault ContributorAchievement: Pull Sharkx2Achievement: Starstruck

Organizations

@WickedOldGames

Block or report maci0

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
maci0/README.md
Marcel Wysocki, Senior Principal Architect @ Red Hat, Singapore

I build AI inference infrastructure and container runtimes, from WGSL compute kernels to distributed GPU clusters, with side quests in binary reverse engineering.

LinkedIn Zig Go Python CUDA Kubernetes

🧠 AI inference, down to the metal

  • agave: LLM inference engine written from scratch in Zig, no Python runtime, no framework tax
  • cwen: Qwen3 inference in pure C: mmap'd Q4 GGUF, AVX-512 kernels, speculative decoding, zero dependencies
  • beam: Ray's API reimplemented in ~1,400 lines of pure Python, enough to run vLLM multi-node. NVIDIA and AMD, validated cross-node
  • vllm-webgpu: vLLM platform plugin targeting WebGPU: WGSL compute kernels instead of CUDA
  • vllm-spark-0731: Serving DeepSeek-V4-Flash on a 2-node DGX Spark cluster: fp8 and nvfp4 MLA kernels, everything measured on real hardware
  • muninn-sidecar: Persistent memory for AI agents, shipped as a Go sidecar
  • gpustack-modelsync: Declarative model placement across GPU cluster nodes
  • toktop: htop for LLM inference engines and the coding agents hammering them
  • gb10-thermal-toolkit: Thermal-driven GPU clock governor for NVIDIA DGX Spark, fixes the under-load power-off
  • gauntlet: Auto-fix review loop, 50 specialized prompts dispatched to whatever AI coding agents you have installed
  • clanker: Self-improving agent harness in Zig, gated self-patch loop with sandboxed WASM tools

🕹️ Reverse engineering

  • rebrew: Compiler-in-the-loop decompilation workbench for binary-matching reversing
  • rebrew-toolchains: Docker images for legacy Windows/DOS compilers, MSVC 1.0-11.0 through Borland, Watcom and Delphi
  • europa1400-networkfix: Fixed multiplayer in a 2001 game the vendor abandoned
  • openmiles: Open reimplementation of the Miles Sound System
  • ct-recomp: Byte-identical decompilation and asset rebuild of Chrono Trigger (SNES)
  • resembl: Assembly code similarity search
  • recoverage: Coverage dashboard for binary-matching decompilation projects
  • europa1400-lua: Lua console injected into a running 2001 game, for reverse engineers and modders

⚙️ Runtimes and clusters

  • katamaran: Live migration for Kata Containers
  • vmetal-openshift: Virtual baremetal OpenShift lab: Redfish BMC emulation, bonded NICs, split-DNS, one Ansible topology

Zig for things that must be fast, Python for things that must exist by Friday.

contribution snake

Pinned Loading

  1. muninn-sidecar muninn-sidecar Public

    Persistent memory sidecar for AI agents, written in Go

    Go 4 1

  2. openmiles openmiles Public

    An open source reimplementation of the Miles Sound System for older games

    Zig 6 1

  3. pelorus pelorus Public

    Forked from dora-metrics/pelorus

    Automate the measurement of organizational behavior

    Python

  4. katamaran katamaran Public

    Live migration for Kata Containers

    Go 1

  5. rebrew rebrew Public

    Compiler-in-the-loop decompilation workbench for binary-matching game reversing.

    Python

  6. gauntlet gauntlet Public

    Run your codebase through the gauntlet: an auto-fix review loop of 50 specialized review prompts dispatched to whatever AI coding agents you have installed.

    Go 1