I write about building AI systems, what I learn from teaching, and the ideas I can't stop thinking about. Technical depth, honest takes, no filler.
Technical
Building Production-Ready RAG Systems: What the Tutorials Don't Tell You
May 2026 12 min read
RAG is easy to demo and hard to ship. After building FabrizioAI and running bootcamps for 1,300+ developers, here are the failure modes nobody warns you about - chunking strategy, retrieval drift, context poisoning, and what "production-ready" actually means.
Action Boundaries in LLM Agents: Why Telling an Agent What Not to Do Is Harder Than It Sounds
Apr 2026 9 min read
When an agent has the capability to execute a task but shouldn't - how do you specify that in a way that's robust, testable, and doesn't break when the prompt changes?
What I Learned Running a Bootcamp for 1,300 People Across Africa
Sep 2025 7 min read
Six weeks. Three timezones. Four guest speakers. One curriculum I wrote from scratch. What worked, what broke, and what surprised me about teaching at scale.
Spring Boot 4 + Java 25: Building a Concurrency-Safe Wallet Engine
Mar 2026 11 min read
Pessimistic locking, idempotency keys, and ACID transactions - a deep dive into how to build a fintech wallet engine that doesn't double-spend under load.
Why Benchmark Performance Is a Bad Proxy for Production Reliability
Feb 2026 6 min read
A model can score 92% on MMLU and still fail your users reliably. Here's why evaluation in research settings systematically lies about what you'll see in production.
Notes on Mathematical Convolution and Its Surprising Applications Beyond Signals
Jan 2025 8 min read
Convolution showed up in my Mathematics degree, image processing, CNN architectures, and RAG retrieval - here's a note on why the operation keeps recurring across disciplines.