10,000+ oil field monitoring stations, each generating 32,000+ data points per sensor. 3-axis accelerometer data coming in raw, needing to be transformed into meaningful orientation metrics (pitch, roll, gravitational forces) so engineers could detect anomalies and do predictive maintenance. No existing pipeline for it.
Wrote C++ algorithms for the sensor orientation math. Accelerometer data in, pitch/roll/gravitational force calculations out. Then built the ETL layer in Python with Pandas to normalize, transform, and aggregate at scale.
On the front end, built a JavaScript visualization dashboard that shows real-time sensor orientation and 3D CAD representations. Stakeholders can look at the dashboard and immediately see which stations are drifting from baseline. The predictive maintenance workflow this enabled was projected to save millions in operational costs.
Delivered and handed off. Pipeline integrated with existing monitoring infrastructure.