Alerting and On-Call for ML Systems
Turn Your Dashboards Into Alerts That Matter: Prometheus Rules for Latency, Errors, and Low Confidence, Routed Through Alertmanager to Slack. (MLOPS: ABSOLUTE BEGINNERS TO PRO IN 100 DAYS - DAY 47)
Sandip Das
Senior Cloud, DevOps, MLOps & ML Platform Engineer | Heading Cloud, DevOps & MLOps for start-ups | AWS Container Hero | Educator | Mentor | Teaching Cloud, DevOps & Programming in Simple Way
Check out the latest posts
Distributed Tracing with OpenTelemetry
Follow One Prediction Request Through Every Service and Step, See Exactly Where the Milliseconds Go, and Link It to Your Logs. (MLOPS: ABSOLUTE BEGINNERS TO PRO IN 100 DAYS - DAY 46)
Structured Logging for ML Services
Replace Print Statements With One JSON Event per Prediction: Inputs, Output, Confidence, Latency, Model Version, and a Request ID You Can Search.
How Hermes Agent Works, and How to Get the Most Out of It
A deep dive into how Hermes Agent by Nous Research works under the hood: one agent loop behind every surface, a tool registry over sandboxed backends, and a learning loop that turns your sessions into memory and skills. Plus 12 habits to get the most out of it.
Custom ML Metrics with Prometheus
Go Beyond Latency: Export Predictions per Class, Confidence Scores, Feature Statistics, and Model Version So Your Dashboard Shows What the Model Is Doing.
Black October 2026: 15 Supply Chain Incidents in 17 Days, and What Every Engineering Team Should Fix This Week
Railway, LiteLLM, Zeabur, OpenCode, OVHcloud, Rank Math, ConnectWise, GhostAction, PHP, Node.js, eMudhra and more. What broke, the real root causes, and a 72-hour, 30-day and 90-day plan to protect your stack.