// /content/posts · registry
data registry_
Notes from the platform layer — CI/CD, observability, supply-chain security, and the long tail of running services no one wants to think about.
#006
2026 · 10
The Agent Says "Done". The Telemetry Has Questions
An AI agent can return 200 OK and still fail the task. A generalized approach to observing agents with OpenTelemetry, the GenAI semantic conventions and Elastic APM: one trace across the service and the agent, tokens and cost, dashboards, evals, and the traps that make panels lie.
16 min →
#005
2026 · 09
An SRE Agent in Two Days: What Our Vibeathon Was Really Like
Artem Chernenko and I joined the AI Agents.PRO vibeathon with a rough idea for an SRE agent. A coding agent gave us a working system quickly. The real work was turning that first draft into something we could test, explain, and trust.
11 min →
#004
2026 · 06
Treating AI Coding Tools Like a Platform Product
Notes on shipping AI assistant skills, agents, hooks, and MCP context as a versioned platform instead of letting everyone run their own prompts. With evals and a couple of numbers that actually moved.
10 min →
#003
2025 · 11
Observability as Code: My Backend Meetup Talk, Now in Writing
A written adaptation of my Backend Meetup talk on observability in microservices — from the three pillars and OpenTelemetry to Terraform-managed alerts and on-call workflows.
6 min →
#002
2025 · 06
Supply Chain Security in Practice: From Container Scanning to Compliance
A practical guide to securing your CI/CD pipeline: container vulnerability scanning, secrets management, SHA pinning, and preparing for compliance audits.
3 min →
#001
2024 · 09
Building a Centralized CI/CD Platform for Microservices
Lessons from centralizing CI/CD workflows into a reusable platform that handles build, test, deploy, and rollback across a microservices organization.
3 min →