TerraSmart PeakYield
Company
Zigron
Role
Senior Full Stack Engineer
Timeline
2024 - 2025
Links
Lower after rebuilding fault detection
Faster fault checks
From solar farms, every day
In short
PeakYield monitors large solar farms, where panels turn to follow the sun, and flags problems before they cost energy.
The problem
The farms send more than a million readings a day. The part that spotted faults was expensive to run and slower than it needed to be.
What I did
I moved fault detection onto a faster, cheaper database setup and rewrote the slowest parts. I also built release notes that update automatically alongside the equipment's software, so nobody has to coordinate them by hand.
The result
- About 50% lower cloud costs
- About 30% faster
- Around 40% fewer user-reported issues over two quarters
Technical detailsFor engineers: how it's built and what it runs on
Approach
Moved the ML and anomaly-detection services from ElasticSearch to PostgreSQL, then rewrote the heaviest queries. Built an automated release-notes system synchronized with firmware versions, and delivered feature launches and bug fixes that cut user-reported issues by about 40% over two quarters.
Architecture
- Node.js/Express services and AWS Lambda endpoints
- Event ingestion streams powered by AWS SQS and SNS queues
- PostgreSQL replacing ElasticSearch for ML and anomaly detection
- Release-notes system synchronized with firmware versions
Highlights
- Real-time monitoring and anomaly detection
- Live telemetry streaming with automated alerts
- Predictive maintenance workflows to reduce downtime
- Operational dashboards for admins and field teams
- Scalable microservices and event-driven pipelines
Stack
Finished reviewing?
Back to Portfolio←