GENKAW
自動化 · AI · 創造 (Automation, AI, Creation)
Ten experiments. Ten lessons. Two trading systems, one life dashboard, one infrastructure stack. Built with Claude AI as a thinking partner.
From curiosity to creation
The Story
Ten projects. Four months. No CS degree. Just logic, curiosity, and an AI thinking partner.
I'm Mito — Support Team Lead by day, builder by night. Everything I build serves one mission: income streams without employer dependency.
A CLI RPG, an enterprise finance platform, a geopolitical simulation, three RL experiments — casting wide to find what sticks.
Stax taught domain-driven design. Ronin taught RL trade-offs on constrained hardware. Rift Codex and WebsiteBuddies taught the difference between impressive and profitable.
Two AI trading systems on real exchanges. A life dashboard tying it all together. Self-hosted infrastructure with zero cloud lock-in. The lessons sharpened into systems that generate revenue.
Current Focus
Life Dashboard
Bathala — project command centre, expense tracking, task management. Multi-user: Mito's ops alongside Iony's business dashboard.
AI Trading Systems
Kraken-LLM and Bitvavo-LLM — AI agents that analyse markets, make decisions, and execute trades on real exchanges. Resuming after Bathala stabilises.
Self-Hosted Infrastructure
Cloudflare tunnels, systemd services, zero trust networking. Own every layer, depend on no one.
What's running right now
The Lab
What's Running
Kraken-LLM
Trading SystemAI-powered crypto trading on Kraken exchange
Bitvavo-LLM
Trading SystemAI-powered crypto trading on Bitvavo exchange
The Toolkit
Python
asyncio
FastAPI
PostgreSQL
DuckDB
Next.js
TypeScript
React
Tailwind CSS
Alpine.js
Claude AI
Claude Code
PyTorch
Reinforcement Learning
Cloudflare
systemd
Docker
Zero Trust
Linux
Lessons from the Archive
Enterprise patterns are powerful. Over-engineering a personal tool is not. Know when to freeze.
Content sites don't scale toward financial independence. High quality, low leverage — wrong economics.
RL on homelab GPU is painfully slow. The gap between experiment and product is wider than expected.
Freelance agencies need clients before code. Building infrastructure before pipeline is building backwards.
Self-hosted email is a nightmare. Pragmatism over purism — use managed services.
Building in public—wins, lessons, and experiments
Updates
Bathala: Life Dashboard & Car Views
Kraken-LLM: Dashboard Overhaul
Kraken-LLM: Agentic Trading Desk
Bathala: Project Command Centre Goes Live
Kraken-LLM: Full LLM Integration
Kraken-LLM: Futures, Risk, and Multi-Strategy
Kraken-LLM: Dashboard & Backtesting Engine
Kraken-LLM: Volatility Breakout Strategy