Developer Profile

Autonomous intelligence meets creative engineering

I build systems where AI agents do real work — and worlds where people want to spend time. Generative AI is not a conversational gimmick to me; it is a substrate for deterministic, goal-oriented execution.

My work sits at the intersection of autonomous agent architecture, serverless edge deployment, and interactive digital worlds. That means designing tool execution engines that behave predictably under failure, memory fabrics that persist across sessions, and pipelines that let a single engineer operate with the capacity of a much larger organisation.

Everything here is built with AI in the loop — not as autocomplete, but as a collaborator with defined tools, reviewable output and its own memory. The blog is a genuine engineering log of that process, including the parts that broke.

What I work on

Autonomous agent architectures

Multi-agent pipelines with goal persistence, state continuation, self-correction loops and structured tool calling — built to finish work, not to chat.

Tool execution & MCP

Deterministic tool routing, Model Context Protocol servers, and integrations that let an agent act on real systems: Cloudflare, GitHub, n8n, databases and messaging.

Persistent memory

Long-term memory fabrics — vector and graph retrieval — so context survives across sessions instead of resetting every conversation.

Edge & serverless

Cloudflare Workers, D1, R2 and Pages. Astro component systems, server-rendered content, and infrastructure that costs almost nothing to run.

Game engineering

High-performance 2D canvas physics loops, celestial gravity attractors, procedural Web Audio synthesis, and browser-native 3D world building.

Automation pipelines

Scheduled jobs, webhook chains and reporting flows — the plumbing that lets one engineer operate with the capacity of a full team.

Projects

Get in touch

Questions about a devlog, the AI workflows, or the games — reach me on Telegram or Discord. I do not publish an email address.