I write and speak about applied AI — the parts where models meet real systems, real data, and real users. Shorter takes live on LinkedIn; the long-form lives here.
A practical guide to my daily agentic workflow — the hardware split, the n8n and Hermes stack, the Claude Code + tmux loop, the boring safety layer, and the rules I follow so agents stay useful instead of dangerous.
Auto-incremented, prefixed, human-readable IDs (INV00001, ORD0003) in Django — four ways, with the trade-offs of each.
PostgreSQL views as first-class Django migrations — fewer surprises, faster reports, and a cleaner ORM surface area.
Turn a webcam into a tripwire with two canvases, a frame diff, and surprisingly little math.
A tiny 2013-era hack, dusted off — using Apps Script as a free, persistent cron for OAuth-flavored side projects.
From vibe coding to systematic engineering. A 2-hour internal training: how AI is changing software development, why context beats prompts, what MCP / LSP actually do, and the working anatomy of agents, skills, and slash commands — with live demos of a full plan → work → critic → compound loop.
Don't compete with AI agents — master them. A workshop on what changes when code production goes nearly free: tech-stack expertise stops being a moat, output stops mattering, and the engineers who win are the Renaissance ones who orchestrate agents across domains.
A short keynote on whether IT — as a department, a discipline, and a profession — still matters in 2025, and what "mattering" looks like once AI eats the routine.