Skip to content
View NAME0x0's full-sized avatar
:dependabot:
I feel like a bot
:dependabot:
I feel like a bot
81 followers · 65 following

Achievements

Achievement: Pull Sharkx2Achievement: QuickdrawAchievement: YOLO

Achievements

Achievement: Pull Sharkx2Achievement: QuickdrawAchievement: YOLO

Highlights

  • Pro

Block or report NAME0x0

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
NAME0x0/README.md

ASCII terminal-style GitHub profile card for NAME0x0 with live stats

NAME0x0 — AI Engineer, Systems Architect, Multi-Agent Wrangler, From-Scratch Absolutist. c/o Muhammad Afsah Mumtaz, Dubai, UAE

Status: open to work — AI / ML / Software Engineering roles, Dubai or remote, UAE Golden Visa (no sponsorship needed), immediate start

Muhammad Afsah Mumtaz (NAME0x0) — AI engineer in Dubai building local-first systems: LLM fine-tuning and evaluation, multi-agent orchestration, and the Rust/C++ tools and shell environments around them. Every claim below links to running code. BSc (Hons) IT, Middlesex University Dubai. Reach me: m.afsah.279@gmail.com

From-scratch manifesto: a car, a language, a programming language, a computer, a mobile phone, clothes — all stamped NOT YET. Excuses: NEVER.

Now building: making 2B models punch above their weight, keeping 11 agents arguing productively about markets

The Work

Pantheon-Trades — 11 LLM agents deliberate every prediction-market trade. Brier 0.149 on a 200-market backtest, 714 Python + 65 Solidity tests, live at pantheon-trades.vercel.app

AVA — fine-tuned a 2B LLM to 82.0% ARC-Challenge / 92.0% ARC-Easy, beating Llama 3.2 3B-Instruct, with a 42 MB QLoRA adapter on a single 4 GB GPU

MALD — native Rust PKM: hybrid full-text + semantic search, cited RAG chat down to exact file and line, fully local, shipped via own Scoop bucket

Also in the workshop: WebDesk (websites and shaders as live Windows wallpapers, C#), OMNI (1.05T-parameter sparse MoE research build, Rust), pane (real Linux desktops on Windows, Rust)

Workshop overflow: WebDesk · OMNI · pane · MAVIS — scriptable Windows shell replacement · Terminus — terminal that ate the desktop · neural-nets — train a network in your browser, six chapters · AGI-Ledger — a permanent record of human belief about AGI

Live Telemetry

Live GitHub statistics for NAME0x0: repositories, stars, followers, commits this year, top languages, and latest shipped commit — regenerated daily from the GitHub API

Shipping log: latest commits pushed across NAME0x0's repositories, newest first, regenerated daily

Consistency report: current contribution streak, longest streak, and total contributions this year

Capabilities

Domain Skills
AI / ML Engineering LoRA/QLoRA fine-tuning, model evaluation & benchmarking, RAG pipelines, multi-agent orchestration, quantized inference (GGUF), model serving — PyTorch, Hugging Face, Ollama
Backend & Data FastAPI, Node.js, REST API design, PostgreSQL, Redis, SQLite FTS5, vector search (HNSW)
Blockchain Solidity, Foundry, on-chain settlement (Arc Testnet), Polymarket CLOB integration
Native Systems & UI Rust and C++ application architecture, Windows shell design, terminal-first UX
Web & Real-Time Next.js, React, Three.js, WebGL, Socket.IO, shader-driven visual systems
Tooling & Delivery Docker, GitHub Actions CI/CD, CMake and Cargo workflows, cross-platform scripting, release packaging

Miyamoto Musashi from Vagabond, sword drawn

Vagabond — c/o Miyamoto Musashi. The way is in training.

Beyond the Repos

  • 🏆 2× hackathon runner-up — Odoo Hackathon Dubai · Curtin University Dubai (Nov 2025)
  • 🪑 Chairman, MDX Computing Society · Committee Chair, BCS Student Chapter
  • 🔬 Occasional open-source research — HAGI

Daily stoic quote, regenerated every morning from a self-hosted quote set

Now playing on Spotify

GitHub contribution graph rendered as a snake eating commits

m.afsah.279@gmail.com · LinkedIn · Portfolio · Pantheon Live

Support me on Ko-fi

BUILT BY HAND. NO TEMPLATE. "YOU MADE IT TO THE BOTTOM."

Pinned Loading

  1. Pantheon-Trades Pantheon-Trades Public

    Eleven AI agents deliberate every prediction-market trade. Restraint is anchored on-chain. Polymarket CLOB + Arc Testnet. 11 LLM providers, one env var. Half-Kelly with caps, self-calibrating, full…

    Python 12 1

  2. AVA AVA Public

    Research and training stack for AVA — a tool-using, memory-aware virtual assistant targeting 4 GB VRAM. Spans custom transformers, verifier-RL, external memory, multi-domain benchmarks, and Gemma 4…

    Python 4

  3. MALD MALD Public

    Native Rust PKM for markdown notes with local AI. Iced desktop GUI, terminal CLI, and TUI — semantic search and cited RAG chat via Ollama. Fully local.

    Rust 4

  4. MAVIS MAVIS Public

    MAVIS is a modern, lightweight, and highly customizable Windows shell replacement built in Rust. It offers a unified desktop environment, integrating a terminal, file explorer, code editor, and sys…

    Rust 3

  5. pane pane Public

    Windows-native Linux environment platform for creating, launching, and supporting real Linux desktops on Windows, starting with Arch.

    Rust 2

  6. AVA — Windows setup notes (Triton + ... AVA — Windows setup notes (Triton + FLA + BitsAndBytes co-existence on Windows 11)
    1 
    # Windows setup notes
    2 
    
              
    
              
    
                3
                
    > Standalone version of the Windows-specific build instructions. Also published as a public gist for easy sharing: **https://gist.github.com/NAME0x0/8fe9084e606d3e7ae17d4f1da6a96667** — keep both copies in sync when editing (`gh gist edit 8fe9084e606d3e7ae17d4f1da6a96667 docs/WINDOWS_SETUP.md`).
    
              
    
              
    
                4
                
    
              
    
              
    
                5
                
    The hard part of running AVA v2 on Windows is **Triton + Flash-Linear-Attention + BitsAndBytes co-existence**. None of them love Windows. This doc captures every workaround that made the released training run work on Windows 11 + RTX A2000 Laptop + Python 3.13 + PyTorch 2.10.0+cu130.
    
              
    
          
    
    
    
  
    

    7.