Cutting LLM costs 60% without dumbing your product down
Caching, routing, and right-sizing models. A practical playbook for keeping an AI feature cheap once real users show up.
Read articleWriting
Notes from building & shipping
Solo-founder SaaS lessons, AI & LLM engineering, and full-stack deep dives — the things I wish I'd read sooner.
AI & LLM$ -60%
Founder{ }
What I learned shipping a multi-tenant SaaS mostly solo
The architecture calls I'd make again, the ones I'd undo, and why "boring" infrastructure was the best decision I made.
Read articleFull-Stack*.tld
Multi-tenant subdomains in Next.js 14 without losing your mind
Middleware, wildcard DNS, and the data model that keeps tenant isolation honest. A field guide from production.
Read articleAI & LLMllama3
Private RAG on your own hardware with Ollama & LLaMA 3
When data can't leave the building: a local retrieval pipeline that's good enough to ship, and where the cloud still wins.
Read articleFounder$$$
Stripe billing for SaaS: the parts the docs quietly skip
Proration, failed payments, plan migrations, and webhooks you actually need. The billing edge cases that bite in month three.
Read articleFull-Stack<T>
Type-safe end to end: the Prisma + TypeScript patterns I trust
How I keep types flowing from the database to the button label — and catch the breakage at compile time, not in support tickets.
Read article