Spark

The challenge

Optima Energy installs connected LED lighting for industrial customers. Every fixture and sensor already reports data, but raw device readings don’t tell a plant manager how much energy and money the retrofit is saving, which sites perform best or how spaces are actually used. Optima wanted a product of its own to sell to those customers, in English and Spanish.

What we built

Spark is a multi-tenant SaaS platform that reads data from the installed lighting and turns it into answers. It only reads: it never controls the lights.

• Ingestion: scheduled jobs pull live status, motion triggers, hourly energy and site topology from the lighting cloud, and store every reading as time series in PostgreSQL with TimescaleDB. Where the source only offers daily totals, Spark computes 15-minute energy itself.
• Savings: kWh, money and CO₂ saved against the pre-retrofit baseline, with everyday equivalents such as trees and car kilometres.
• Portfolio view: a world map of sites, efficiency rankings by kWh per m², occupancy estimated from motion sensors, and statistics by day, week or month.
• Digital twin: 2D and 3D views of each site, built from the lighting design files.
• AI summaries: a daily briefing in English and Spanish, as text and voice.
• For teams: invite-only accounts with two-factor login, roles for Optima and each customer, CSV export, a REST API with tokens and an audit log.

Results

Spark went from first commit in April 2026 to production on two domains, one in English and one in Spanish. The codebase runs with full test coverage and strict static analysis, so new features ship without fear of breaking the numbers customers rely on.