Kuppa — Home Energy Analytics
Scaled a live UK home-energy platform that turns a detailed home profile — structure, materials, occupants, vehicles, appliances — into an AI-generated efficiency report, shipping new features, analytics, and reporting.
- Stack
- Node.js · Express · PostgreSQL · Redis · Docker
The challenge
Kuppa helps UK homeowners understand — and cut — their home's energy consumption. A user describes their home in detail: floors, walls, roof, construction materials, household size, vehicles, electrical appliances. The platform turns that profile into an AI-generated energy report with targeted recommendations for reducing consumption.
I took over a live distributed system I hadn't designed — and the mandate was to extend it significantly without destabilising what users already depended on.
What I engineered
- New product features built into the existing distributed service architecture, engineered to fit its conventions rather than fight them.
- An analytics layer across homes, inputs, and recommendations — giving the business real visibility into how the product is used.
- Advanced reporting on energy profiles and recommendations, for both end users and the admin side.
Architecture
Node/Express services → PostgreSQL (Sequelize) holding home profiles, inputs, and generated reports → Redis caching → Docker deployment. The AI report pipeline consumes the structured home profile and returns recommendations that the backend stores and serves.
Impact
- Major features shipped into a live production system with zero disruption
- A new analytics capability the business didn't have before
- Richer, more actionable energy reports for UK homeowners
What I took from it
Inheriting a production codebase is a different discipline from starting one. The work was reading before writing — understanding why every decision was made, then extending the system in a way its original authors would recognise.