Grid operator, Central Europe. We collect sensor data from 8,000+ substations at 1-minute resolution. Current use: manually reviewed by engineers who look for anomalies.
We want two things:
1. Anomaly detection — flag readings that suggest equipment fault before it causes an outage
2. Short-term load forecasting — 15-minute and 1-hour-ahead predictions at substation level
Both need to handle missing data (sensors go offline), handle the periodicity of power consumption (daily/weekly/seasonal patterns), and be explainable enough that our grid engineers trust the output.
Huge dataset, real operational stakes. If you've worked with power system or utility data before, lead with that.