Back to projects
AIPythonSpark

Project Helios

Intelligent Data Pipeline

Project Helios - Intelligent Data Pipeline

Helios processes 2 TB+ of telemetry daily using a distributed Spark cluster,
feeding real-time anomaly detection models that underpin our SLA guarantees.

Key outcomes:
  • P99 pipeline latency: 11 s
  • Anomaly detection accuracy: 97.4%
  • Saved clients $4M+ in avoided incidents (2022)

Stack: Python · Apache Spark · Kafka · MLflow · PostgreSQL
Year
2021
File
ai-pipeline.html