Building the Sea Link in 3D
From Python-generated bridge geometry to an animated bus ride, camera rigs, sunset lighting and water reflections in WebGL.
Writing on AI, code and teaching
Notes from my projects, alongside guides to reviewing AI-generated code, choosing an AI pilot and planning training.
Articles and guides
Putting a photo on the Sea Link in WebGL
A technical look at mapping a personal photo onto a 3D bridge, reflecting it in water and saving the result.
Prove You Know Me: A Friendship Quiz in Nine Squares
Building a friendship quiz where the creator sets the answers and AI helps choose the options.
A Review Checklist for AI-Generated Code
Work through a fictional support-transcript feature, then copy the checklist into your next pull request.
AI Keynote or Workshop: Which Does Your Team Need?
Compare preparation, participation and takeaways, then use the worksheet to choose a format for your audience.
How to Choose a First AI Pilot
Define the task, data limits, human review and stop condition before testing a pilot.
How to Brief an AI Trainer
Give a trainer the participant details, tool access and delivery conditions they need to plan the exercises.
Published elsewhere
A few articles from my time at Pieces. Product and model details reflect their original publication dates.
Back to school: the best AI programming languages to learn first on Pieces
Choosing a language for AI work means thinking about experimentation, deployment and the people who will maintain it. A practical look at those tradeoffs.
What are the best Rewind AI alternatives to consider in 2025? on Pieces
Memory tools through a developer’s eyes: local processing, code context, integrations, and the difference between recording activity and saving something useful.
Bigger is not always better: comparing LLMs and SLMs on Pieces
What smaller models offer when API costs and latency matter. Lessons from exploring where broad capability helps and where a focused model fits.
Exploring Microsoft’s Phi-3-Mini and its integration with tools like Ollama and Pieces on Pieces
Running Phi-3 locally with Ollama and organizing generated snippets in Pieces, with practical setup steps and a look at the model’s context limitations.