annias

Hi there 👋

Charles “Annias” Macnish

AI Systems Architect · Creative Technologist · Systems Builder

I design and build AI systems, agent infrastructure, automation, developer tools, interactive software, VR/XR projects, games, audio systems, and generative media.

My background crosses software development, IT architecture, electronic music, realtime media, virtual worlds, automation, and applied AI. I tend to work across the full system instead of staying inside one layer, from infrastructure and orchestration through interfaces, creative tooling, and the environments people actually use.

Current work is heavily focused on autonomous and semi-autonomous software systems, especially agent orchestration, memory, context management, tool use, local compute, multimodal interfaces, and the infrastructure required to connect those pieces reliably.

🧠 AION CANNON

AION CANNON is my primary AI systems architecture project.

It is a modular environment for connecting reasoning models, specialized agents, tools, memory systems, applications, local infrastructure, creative systems, and human workflows.

                         HUMAN
                           │
                           ▼
                    ┌─────────────┐
                    │   AGENTS    │
                    └──────┬──────┘
                           │
             ┌─────────────┼─────────────┐
             ▼             ▼             ▼
          MODELS         TOOLS         MEMORY
             │             │             │
             └─────────────┼─────────────┘
                           ▼
                 ORCHESTRATION LAYER
                           │
          ┌────────────────┼────────────────┐
          ▼                ▼                ▼
      SOFTWARE          SERVICES      INTERACTIVE
                                        SYSTEMS

The work currently covers areas including:

Agent orchestration and delegation
Model Context Protocol infrastructure
Persistent and cross-session memory
Context routing, retrieval, and compression
Local and hybrid model infrastructure
Multi-agent development workflows
Browser and computer-use agents
Voice and multimodal interfaces
Knowledge systems
Tool and application integration
Agent evaluation and observability
AI-assisted software and game development
Human and agent collaboration

The main AION CANNON system is private. A number of supporting experiments, integrations, modified upstream projects, and research branches are public on this account.

🔬 This GitHub Is a Workshop

This account is not meant to be a collection of polished demo applications.

I use GitHub heavily for research, experimentation, integration work, prototypes, modified upstream projects, reference implementations, and software that eventually becomes part of larger systems.

A fairly normal development path for me looks like this:

research
   ↓
prototype
   ↓
reverse engineer
   ↓
modify
   ↓
test
   ↓
connect
   ↓
instrument
   ↓
evaluate
   ↓
reuse what works

You will find original software alongside forks and experimental branches.

A fork on this account may exist because I am studying it, modifying it, testing integration boundaries, exposing it to agents, adapting it to another environment, or using it as one component in a larger architecture.

I am gradually adding more documentation around those relationships where it is useful.

🤖 AI Systems & Agent Infrastructure

Most of my current software work deals with the systems surrounding foundation models.

That includes:

Agent runtimes
Tool calling
MCP servers and clients
Multi-agent coordination
Persistent memory
Semantic retrieval
Knowledge architecture
Context compression
Model routing
Local inference
Browser automation
Computer use
Voice interfaces
Agent-driven development
Evaluation and telemetry
Human approval and control layers

I am particularly interested in what changes when agents are given persistent environments, usable memory, proper tools, observable state, and the ability to operate software directly.

🎮 Games, Simulation & Agent Development

Game development is one of the environments where I test agent systems.

I am interested in agents that can participate in the actual development loop:

inspect
   ↓
edit
   ↓
build
   ↓
run
   ↓
observe
   ↓
interact
   ↓
debug
   ↓
iterate

That includes work with browser-driven game environments, Spawn.co, Unity, procedural systems, simulation, collaborative agent development, and tools that allow models to inspect and manipulate running applications.

Games are useful here because they combine code, persistent state, graphics, physics, interaction, systems design, debugging, and subjective evaluation in the same environment.

🥽 VR, Spatial Computing & Virtual Worlds

I have been involved with VR and virtual worlds for years, particularly VRChat and the surrounding creator ecosystem.

My work and experiments have included:

VRChat worlds and tooling
Unity
OpenVR and related runtimes
VR application integration
OSC
Avatars
Social VR
Realtime media
Procedural environments
Non-Euclidean environments
Spatial interfaces
Virtual performance
Agent integration with interactive environments

I am interested in VR both as a software platform and as a social environment with its own behaviors, communities, creative practices, and interface problems.

🎛️ Audio, Music Technology & Generative Media

I have been producing electronic music and DJing since 1997.

That background has led to a parallel body of technical work involving:

Digital signal processing
Audio analysis
Source separation
Generative composition
Neural audio
Sampling and manipulation
Procedural sound
Audio-reactive graphics
Realtime audiovisual systems
Music production tools
Generative image and video systems

A lot of my creative software exists because I wanted a tool that did not already exist, or because I wanted existing tools to interact in ways they were not originally designed to.

🛠️ Technologies

I use whatever stack makes sense for the system.

Languages

Python · TypeScript · JavaScript · C# · Lua · SQL

AI & Agent Systems

LLMs · MCP · RAG · Embeddings · Vector Search · Local Models · Tool Use · Agent Orchestration · Multimodal AI

Application & Infrastructure

Node.js · Docker · REST APIs · WebSockets · Playwright · Git · GitHub · Azure · Microsoft 365

Interactive & Creative

Unity · VRChat SDK · OpenVR · OSC · WebGL · WebGPU · Blender · Audio DSP

🧩 Areas I Keep Coming Back To

AI agents
Systems architecture
Developer tooling
Automation
Knowledge and memory systems
Game systems
VR and spatial computing
Creative technology
Audio and DSP
Generative media
Realtime systems
Human-computer interaction
Simulation

There is a common thread through most of it: understanding how different systems work, finding useful connection points between them, and building the infrastructure required to make those connections practical.

🌎 Earthside Collective

A lot of my technical and creative work also lives under Earthside Collective, a group of artists, developers, musicians, VR creators, game developers, and other builders working on independent and collaborative projects.

Earthside covers software and AI systems, consulting, web and application development, music, visual media, VR, games, live production, and experimental creative technology.

🔗 Art, Work & Socials

Build weird things. Understand how they work. Connect the useful parts.

Navigation Menu