STATUS · LINKING
High-throughput ingestion and analytics platform for distributed solar telemetry. Sub-second query at 1.2B events/day.
Solar fleet generated 1.2B telemetry events per day. Existing data lake ran 90s+ queries and crashed at peak.
Time-series database with rolling materialized views, columnar storage, and a thin GraphQL layer for the customer-facing console.
Sub-second median query latency at 1.2B events/day. Cost per event dropped 78% versus the prior stack.
We built a co-pilot that classifies CI/CD failures, generates fix candidates, and coordinates remediation across GitLab and GitHub.
Real-time route optimization across 1,200 vehicles. Replaced an ILP solver and a half-time analyst.
Document understanding pipeline for 14 clinical specialties. From intake form to structured EHR record in 90 seconds.