Using AI coding agents without wrecking your codebase is where most people are struggling right now — that's my sharpest area. Behind it: 20+ years of production Python, Linux, and system design, two CTO roles, and a Master's in Data Analytics with a machine-learning focus. I currently build and run production infrastructure at a fintech.
I'm most useful to you on three things:
• AI-assisted development done right — getting real speed out of coding agents while keeping your codebase coherent, reviewable, and maintainable.
• Production Python & system design — architecture review, design-interview prep, "why is this slow / fragile / hard to change," and how a senior engineer actually reasons about a system.
• Linux, Docker & data infrastructure — real production ops, not toy examples. Debugging, deployment, storage, scaling.
I teach with a bias toward the durable stuff: model your data and your domain well, and most of the hard problems get easier. It's also why I keep a standing interest in semantic web standards — RDF, JSON-LD, SHACL, knowledge graphs — and why I'll push you to get the model right before you write the code. Patient, clear, and I'll make sure you actually understand the why, not just the fix.
A good first session — Agent-Readiness Architecture Review (60 min): bring a real repository and we'll find the architectural assumptions that exist only as tribal knowledge, the invariants worth making executable, and where agent-driven change is most likely to cause drift.
