Home/Work/The Lab

Case study · Independent R&D

The Lab

Self-funded engineering that keeps client work ahead of the curve — an AI trading system, an EOS business platform and multi-agent AI tooling. The deep end, getting deeper. Every experiment here becomes capability your project can use tomorrow.

  • Independent builds
  • Ongoing
  • Python · TensorFlow
  • Next.js · Convex · Claude
01

Why a lab exists

Client work rewards reliability; growth demands risk. The Lab is where the risk happens — self-funded projects hard enough to force real learning, so that by the time a client needs machine learning, real-time sync or multi-agent AI, it is not the first time I have shipped it.

02

The AI trading system

A machine-learning trading system built end to end — data pipelines, models, backtesting and risk management.

  • Python 3.10 with TensorFlow and Keras LSTM models; pandas and NumPy pipelines
  • RSI, MACD, ATR and Bollinger Bands across five timeframes — one minute to four hours — on SOL-USDT spot (OKX)
  • A dedicated backtesting framework, drawdown limits, Kelly-based position gating and trailing stops
  • The hardest lesson engineered in: knowing when not to trade
03

Aurora

An EOS business platform concept — the operating system I have run companies on, distilled into focused software.

  • Scorecards, accountability charts, meetings and priorities in one focused system
  • Next.js, TypeScript, Convex, Tailwind and shadcn/ui; offline-first PWA with real-time sync
  • Designed against hard targets: over 90% internal adoption and 25% faster task cycles
04

And beyond

  • Multi-agent AI desktop tooling, Blender scripting and Unreal Engine 5 experiments
  • One hundred interactive engineering exhibits, built by hand and running live on this site

The proof does not live in a document — it runs in your browser. The Lab page holds one hundred hand-built interactive exhibits, and the Applied page maps each technique to the industries where it earns money.

05

The outcome

A standing library of proven capability — machine learning, real-time platforms, physics, visualisation and multi-agent AI — that client projects draw on directly. When a proposal says "this is possible", it is because a version of it already runs in the Lab.

The kaizen part: every experiment here becomes capability your project can use tomorrow.

Python · TensorFlow · Next.js · TypeScript · Convex · Claude

Your project next

Put the deep end to work.

If your project needs more than a brochure — AI, automation, real-time systems — this is the studio that has already built them.

← Previous: D AND P Branding Next: Seventh Star →