LabExperiments

Things built to find out whether they'd work

No client, no deadline, no requirement to be useful. Some of these answered the question and stopped there — which was the point.

K-7
EXP-01open source

D3 Advanced Visuals

An enterprise analytics platform built in the open: reusable chart components that survive contact with real data and two themes.

  • D3.js
  • Next.js
  • TypeScript
EXP-02prototype

Handwritten Text Recognition

An OCR pipeline where most of the work happens before the model does. Preprocessing decides what recognition even gets a chance at.

  • OpenCV
  • Tesseract
  • Python
EXP-03prototype

PUJ Obstruction Detection

A deep-learning pipeline over ultrasound data. Medical imaging punishes confident models with thin evidence.

  • Python
  • TensorFlow
  • Deep learning
EXP-04prototype

Carbon Footprint Calculator

Electricity-based emissions estimation, with offsets visualised: a number nobody feels, made into something they can see.

  • JavaScript
  • Data visualisation
Next · MissionsEvery one started as somebody's bad afternoonThe experiments above answered a question. These ones had to survive contact with users.