Blog
Architecture, AI engineering and the decisions that actually determine whether a product ships and holds up.
Most MVPs fail before launch because they try to prove too many things at once. Here is the scoping method I use to get a fundable product live in 6-10 weeks.
Aug 12, 2026 3 min read
Most architecture mistakes are recoverable with a refactor. A handful of database decisions are not. Here are the ones worth getting right on day one.
Aug 6, 2026 3 min read
Retrieval-augmented generation gets pitched as the fix for AI hallucination. It helps — but the reliability comes from evaluation and guardrails you build around it, not the retrieval step itself.
Jul 31, 2026 3 min read
Cron jobs are the easiest way to ship background work and the easiest way to get paged at 3am. Here's how to design job processing that fails safely.
Jul 24, 2026 3 min read
"We'll fix it later" is not a plan, it's a deferred invoice. Here's how I put a real number on technical debt so it competes fairly against new features for a place on the roadmap.
Jul 16, 2026 3 min read
URL versioning, header versioning, no versioning at all — the strategy matters less than whether you've decided what a breaking change actually is before your first external integration.
Jul 8, 2026 3 min read