From the build log
Concrete technical posts on Agentic Ops, Kubernetes platform engineering, MLOps, and AI FinOps. We write about what we actually ship.
Why We Run Four Kubernetes Scanners Simultaneously (and Score Them)
Pluto catches deprecated APIs. Polaris scores policy compliance. kube-score grades reliability. OSV.dev surfaces CVEs. No single tool is enough — here's how we combine them into one ShrikeOps score.
The $847K GPU Waste Problem — and the Math to Fix It
Most AI teams over-provision inference nodes by 40–60%. We built a mathematical rightsizing model that only executes VM resizes when projected SLO compliance stays ≥95%. Here's the formula.
AlloyDB pgvector for Production RAG: RLS, Sub-100ms Retrieval, Multi-tenancy
We run AlloyDB with Row-Level Security keyed on tenant_id for every Warble Cloud customer. Here's our schema, index configuration, and how we hit sub-100ms retrieval on 10K+ remediation recipes.
ShrikeOps Deep-Dive: MCP + Kubernetes
Model Context Protocol opens a new primitive for AI-to-infrastructure communication. We explore how MCP bridges Gemini 2.5 Flash to live cluster state — and what it means for the future of Agentic Ops.
Sovereign AI: Passing FinReg Audits in 48 Hours with VPC Service Controls
JPMorgan and BNY-grade compliance isn't optional for enterprise AI. We walk through the exact GCP architecture — VPC Service Controls, Identity Platform, RLS — that lets us pass security reviews in under 48 hours.
Get posts in your inbox
One email per post. No summaries, no digests — just the full technical post when it ships.
Subscribe via contact form