A day in the life

Writing on enterprise architecture, cybersecurity, infrastructure, observability, and applied AI.


PromQL Optimisation Patterns for High-Cardinality Environments

Concrete PromQL optimisation techniques for platform engineers operating at high cardinality: label matchers, recording rules, histogram strategies, aggregation ordering, and failure modes that silently degrade query performance in production.

14 min read   ·   2026

Claude Code as an Orchestration Layer Across Multi-System Enterprise Environments

How Claude Code's sub-agent model, hook system, MCP integration, and worktree isolation compose into a production-grade orchestration layer for multi-system enterprise workflows — with architecture patterns, failure modes, and a practical implementation checklist.

28 min read   ·   2026

LaTeX Masterclass Week 1: What Is LaTeX and Why It Matters

A clear, honest introduction to LaTeX for researchers, PhD students, and technical writers: what it is, how it differs from Word and Google Docs, why typesetting quality matters, and when LaTeX is the right tool — and when it isn't.

13 min read   ·   2026

Security and Credential Management for AI Agents with Filesystem and API Access

Concrete attack vectors, secrets management integration, filesystem sandboxing, agent identity patterns, and the production failure modes that credential-bearing AI agents introduce — written for engineers building or operating them.

30 min read   ·   2026

TikZ and pgfplots: Publication-Quality Figures That Don't Break Your LaTeX Build

From first principles to production-ready diagrams: how to use TikZ and pgfplots correctly for journal and conference figures, including coordinate systems, styling architecture, common failure modes, and externalization for large documents.

14 min read   ·   2026

OpenTelemetry Migration in Production: Trade-offs Every Platform Team Gets Wrong

Concrete collector topology decisions, SDK migration patterns, tail sampling configuration, cardinality budget management, and rollback architecture for teams migrating from proprietary agents to OpenTelemetry in Kubernetes production environments.

19 min read   ·   2026

MCP Server Architecture and Tool Chaining in Production Agentic Workflows

Beyond the demo — concrete patterns for MCP server design, tool chaining economics, idempotency, and the architectural trade-offs that determine whether an agentic workflow survives production.

10 min read   ·   2026