-7c5cff?style=for-the-badge){kind=icon}
-22d3ee?style=for-the-badge){kind=icon}
CirqHub
An AI-driven circular economy platform for physical market vendors, built with Base44 as an AI-assisted development platform.
01 The problem
Physical market vendors need a fast way to decide what happens to unsold surplus before it goes to waste — donate, barter or reuse — and how to plan staffing and layout for the day ahead. The constraint: four days, a team, and a demo at the end.
02 Scope & plan
Specification came before implementation: MRD, PRD, system architecture, database schema and REST API design. I then planned the four days day by day against a fixed deliverable list — eight-plus artifacts, from market research to the pitch deck — so the team always knew what "done today" meant.
03 My role
Team member owning planning and the spec: I co-designed the architecture, directed the AI-assisted implementation (Base44) against the spec, and ran QA and code review before sign-off.
04 Key decision & risk
An AI recommending what to do with food is a safety risk, so the controls went into the specification rather than being left to the build: deterministic calculations kept separate from LLM-generated text, a database-enforced allow-list of 12 external data sources, and hard-coded food-safety rules. QA then checked each against the spec before sign-off.
05 Outcome
Delivered on schedule: all eight-plus artifacts plus a working end-to-end demo, with vendor decisions, the operator capacity view and offline/online sync functioning together.