Over two decades of technical decisions under incomplete information and unclear constraints. Understand the problem, surface the trade-offs that matter, make the call that keeps risk manageable.
Co-founder of SoftInstigate and creator of RESTHeart, in production for 10+ years.
How I work
Unclear systems and constraints
My work usually starts where the problem is still forming. Requirements are incomplete, stakeholders disagree on priorities, constraints are implicit, and the system is already running in production.
Decisions under pressure
Systems break when assumptions collide with production reality. The decisions that matter are the ones made with incomplete information, where the consequences of being wrong are visible and expensive.
Long-term consequences
Architecture decisions are difficult to undo. A choice that saves a month now can cost a year later. Each one carries the question: how hard will it be to change course if the assumptions were wrong?
AI adoption and architecture
AI is a multiplier of what is already there. It can accelerate development, and it can accelerate decay.
AI follows a pattern: Envision, Experiment, Launch, Scale. Most startups and scale-ups are between Experiment and Launch. They have something working, but the architecture was built for speed, not for the load AI puts on it.
The transition from pilot to production is where decisions become expensive to undo: what happens when the model changes, who owns output quality, how do you handle costs, what is the fallback when the AI is wrong.
Open source
RESTHeart: Agent-ready backend for MongoDB
Facet: Data-driven web framework: turn APIs into HTML, zero code
Ermes Mail: Async e-mail library and CLI for Java
All projects on GitHub →
Teaching
Created and delivered technical courses and workshops for enterprise clients and partner networks across Italy and EMEA. Teaching forced me to articulate why certain decisions are made, which sharpened my own judgment.