Skip to content
View apollo-mg's full-sized avatar
  • Self Employed
  • Near Indianapolis Indiana

Block or report apollo-mg

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
apollo-mg/README.md

Profile README


```markdown

Hi, I'm Mark. 🛠️


I am an AI Systems Architect focused on bare-metal infrastructure, constrained hardware optimization, and edge-deployed intelligence. 
After 14 years of building and running traditional businesses, I transitioned into systems engineering with a specific focus on Local LLM Orchestration and Deterministic Architectures. I don't build API wrappers; I build the plumbing that makes complex models run reliably in air-gapped, zero-trust environments. 
### What I Build My current primary focus is Project Apollo—a localized, multi-agent AI framework designed to run entirely on consumer-grade hardware (AMD/ROCm).

  • VRAM Engineering: Designing strict memory-management protocols to hot-swap 30B+ MoE models within a strict 16GB ceiling.
  • Cascading Inference Routers: Building "Three-Mind" architectural logic (0.6B Gatekeeper -> 8B Thinker -> 30B Architect) to optimize inference cost and latency.
  • Air-Gapped RAG Pipelines: Developing automated "Librarian" scripts to chunk, hash, and ingest WORM data into local vector databases for completely private, semantic retrieval. 
### My Engineering Philosophy
  • Architecture over Syntax: I leverage AI to generate syntax while I provide the system design, constraints, and business logic.
  • Pragmatism over Hype: I focus on the "Post-Bubble" reality of AI—reducing runaway cloud compute costs by replacing insecure API dependencies with optimized, sovereign, open-weight models.
  • Red Team Thinking: Designing for zero-trust, air-gapped security rather than relying on external corporate guardrails. 
### Current Tech Stack
  • Hardware: AMD RDNA 4 (RX 9070 XT), evaluating RISC-V / Tenstorrent Scale-Out architectures.
  • Software: Ubuntu, Native ROCm, PyTorch, vLLM, WORM Databases (ChromaDB), Python. 
### Let's Connect I'm actively seeking roles in Aerospace, Defense, Advanced Manufacturing, or infrastructure teams where bare-metal execution and air-gapped security are baseline requirements, not afterthoughts. 
* Based in Indianapolis, IN.
  • Open to Hybrid/On-Site roles and roles requiring clearance. ``

Popular repositories Loading

  1. ComfyUI-Zluda-MyFix ComfyUI-Zluda-MyFix Public

    My working ComfyUI ZLUDA

    Python 2 1

  2. RDNA4-ComfyUI-Patches RDNA4-ComfyUI-Patches Public

    Python 1

  3. urban-spork urban-spork Public

  4. Project-Apollo Project-Apollo Public

    An air-gapped, local MoE swarm and routing architecture designed for absolute human agency.

    C++

  5. apollo-mg apollo-mg Public

    Professional/Personal Informationb

  6. open-multi-agent open-multi-agent Public

    Forked from JackChen-me/open-multi-agent

    TypeScript multi-agent orchestration engine — one runTeam() call from goal to result. Multi-model teams, auto task decomposition, parallel execution. 3 runtime dependencies.

    TypeScript