A Healthcare and Life Sciences Company Rebuilds Kubernetes Monitoring With ML-Based AIOps

Project context

Monitoring had grown across two incompatible setups that platform teams could no longer trust. SquareShift rebuilt it on what the client actually runs, adding log-noise tagging and ML anomaly detection.

Project objectives

Rebuild dashboards on Elastic Agent data.

Reduce log noise safely and replace fixed thresholds with anomaly detection.

Solution delivery

SquareShift moved 207 dashboard panels from OpenTelemetry to Elastic Agent, matching the originals across five connected staging clusters.

It built a log pipeline that tags noise and keeps errors, and tested it on 934,077 real log records with zero errors affected.

It added seven ML detectors for crash loops and memory leaks, moved alert rules into Terraform, and handed over documentation.

The outcome

The rebuilt monitoring matches the originals across five connected staging clusters. It was tested on 934,077 real log records, with zero errors affected.