E-commerce experiments
Storefront and platform experiments run to understand conversion, merchandising and the mechanics of selling online first-hand rather than from a case study.
Independent Product Lab
Built after work,paid for in advance
Real products, shipped independently, funded by part-time warehouse and forklift shifts rather than by a salary or an investor. This is where the learning compounds.
The cycle
Six steps that keep repeating. The first one is the reason the other five are possible.
- 01
Earn
Part-time warehouse and forklift shifts around study and full-time work.
- 02
Invest
Into certifications, subscriptions, AI API credits, cloud hosting and domains.
- 03
Build
Prototypes, storefronts and full-stack products, built end to end.
- 04
Launch
Into production, where real users and real constraints apply.
- 05
Learn
From what breaks — deployment, cost, reliability, and what nobody uses.
- 06
Improve
Feed it back in. The next build starts further along than the last.
What came out of it
This portfolio
Next.js, TypeScript, Three.js and a retrieval-based AI assistant — built as a product rather than a page, with its own content architecture and deployment pipeline.
Experimental prototypes
Smaller AI workflow prototypes built to test an idea quickly. Most are never shown to anyone; that is what makes them cheap enough to be useful.
What building alone teaches
Ship it, or it did not happen
A prototype that never reaches a user teaches you about building. Getting it into production teaches you about hosting, cost, credentials, failure modes and the gap between working and reliable.
Constraint is the point
A budget that comes from shifts rather than a salary forces genuine prioritisation. You do not add a service because it might be useful when you are personally paying for it monthly.
Build the thing you can maintain
Independent products have no team behind them. Every dependency, every abstraction and every clever decision is one you will be maintaining alone at some point.