Loading…
Practical lessons from shipping real products — architecture, AI, performance, and the craft of engineering.
Coding agents went from autocomplete to teammates in under a year. Here's how I actually delegate work to them without losing control of the codebase.
"Vibe coding" — building by describing what you want and accepting what the AI gives back — is incredible for prototypes and dangerous for production. Knowing the line is the whole skill.
AI can write production code that's genuinely excellent — and production code that's confidently, dangerously wrong. An honest accounting of both, and how to keep the good.
Clever prompts were never the real lever. The engineers getting great results from AI are the ones who master what the model can see — not how they phrase the ask.
AI slashed the cost of writing code. It did nothing to lower the cost of owning it. That gap is where the next decade of technical debt is being quietly created.
When AI writes most of the code, review stops being a formality and becomes the most important engineering activity you have. It also needs to change shape.
Most AI agent demos never make it to production. Here's the architecture I use to build grounded, reliable agents that businesses trust.
Speed is a feature. Here's how I keep Lighthouse scores in the high 90s without sacrificing rich, animated experiences.
Retrofitting multi-tenancy is painful. Here's how to model tenants, data isolation, and billing so your SaaS scales cleanly.
Have a product in mind? Let's turn it into something users love — fast, scalable, and beautifully engineered.