I build the full stack — systems, languages, and developer tools — almost entirely in Rust, almost entirely from scratch.

An operating system. A text editor. An AI coding assistant. A blockchain. A borrow checker. An agent orchestrator. Neural networks. Programming languages. Simulators for synthetic life.

Most of it is open source; the commercial products and services below help me keep building in public.

Personal stretch goal: Build every tool I rely on — nothing I use should be something I couldn't have written myself. Full ownership of the stack.

I give all of this away for free as a shoutout to all the known and unknown open source hackers out there who influenced me, either directly or indirectly.

I run Vibe Coding Agency — shipping production AI infrastructure, agentic systems, and GPU cloud access for teams that need to move fast.

These three represent my core thesis: if you use it every day, you should own it.

Project	What	Why It Exists
RegicideOS	Foundation OS for higher-layer AI products — Gentoo + BtrFS + COSMIC Desktop, every layer optimized for ML. Variants: RegicideOSArch (available now) and musl-based builds (future)	Because an OS that doesn't understand GPUs is obsolete
openlibertas	Terminal-based AI coding assistant with multi-provider, MCP tools, multi-agent delegation, voice mode	Because your coding assistant should be yours — free forever, no subscriptions
neoxide	Neovim-inspired terminal editor with LSP, tree-sitter, Lua plugins, splits, registers, macros, marks	Because every developer deserves an editor that matches how they think

A curated set of projects that power my commercial work and research. See the full awdemos org for experiments in blockchain, languages, simulations, and games.

These projects form the core of how I build, orchestrate, and reason with AI agents — from terminal assistants to distributed multi-agent systems.

Not everything is scratch-built. Some things are battle-tested patterns you can actually use in production:

• Kubernetes — 100+ deployment patterns, Cilium eBPF, multi-cluster federation
• LLM Infrastructure — RAG pipelines, fine-tuning, multimodal benchmarks
• KV Cache Infrastructure — Generate, quantize, and host KV caches for LLMs with kvcdn-cli + KV Cache Store
• CI/CD — Dagger Go pipelines, container-native builds, security scanning
• IaC — Pulumi (Go/TypeScript) for Azure & AWS multi-tenant infrastructure
• Rust — Matrix ops, CLI boilerplate, async patterns that don't leak memory

git clone https://github.com/awdemos/demos.git

Reproducible. Observable. Composable.

I design systems from first principles. Small tools that do one thing well and compose cleanly — the Unix way, but for the AI era.

• I don't trust black boxes. If I can't audit it, I don't ship it.
• I prefer code over YAML. If it can be code, it should be code.
• I believe in AI sovereignty. Models you own, data you control, infrastructure you can audit.
• I value memory safety. Not because Rust told me to, but because debugging segfaults at 3am is a waste of a life.
• I'm a huge fan of tried and true principles and underappreciated technologies. I can work on everything from a PDP-11 to FreeBSD and post-Linux kernels with equal fluency.

My formative years were spent on Slashdot and deep in open source culture — that ethos of transparency, collaboration, and building in public is still how I work today.

What Kimi 2.7 has to say about my coding style: A technically deep, curiosity-driven engineer with strong systems programming instincts, a Rust-centric toolchain, and heavy involvement in AI/automation tooling. Likely experiments broadly and iterates fast rather than shipping polished products. Possible backgrounds: systems engineer, game dev, AI/ML tooling, or indie hacker.

• Best-in-class infrastructure engineering
• The fastest CI/CD on the planet
• Extreme performance optimization
• Extreme Linux and open source security practices

A lot of this work is anchored in one idea: the future of intelligence is unified, not fragmented.

I'm exploring what it means to build AI Pangaea — the convergence of distributed intelligence islands into a more cohesive, secure, and capable continuum:

• Superintelligent entities with persistent memory and recursive self-improvement
• Security at every layer — sandboxing, guardrails, consensus verification, confidential execution
• Memory & continuity across sessions, devices, and intelligence islands

📄 AI Pangaea: Unifying Intelligence Islands — arXiv:2509.17460

• Vibe Coding Agency — products, consulting, and the Open Superintelligence Stack
• Book a call
• LinkedIn
• Twitter/X
• andrew@awdemos.com
• Enterprise AI Atlas

If you're working on something in the intersection of systems, AI, and security, I'd genuinely love to hear about it.