Archive · Five pillars

Considered writing.

Long-form essays, published when ready. Five pillars chosen for coherence with twenty years of work across tech management, software architecture, cloud, and AI. Not courses, not motivation, not anything you can find in feeds.

A

Tech management

Leadership inside Italian companies

2 essays

  1. The trust you receive is a debt

    When someone trusts you before the proof, the risk passes from their hands to yours. On the weight of trust received, and the doubt that never closes. Second of two pieces on trust.

    ·7 min read

  2. Trust doesn't arrive with you

    When you join a new team, you assume your good faith arrives before you do. It doesn't. On building trust in a place that decided, before it met you, that you would fail, and why only a price you pay yourself moves anyone who has been burned before.

    ·5 min read
B

Software architecture

Web, systems, and design (see Two Webs)

3 essays

  1. Siebel already did it

    In 2006 I was programming inside Siebel, a platform that shipped by default the architectural patterns we now chase with three separate tools. On what gets lost when we rebuild from scratch or let an AI generate the code: not the pattern, but the memory of the problem it solved.

    ·6 min read

  2. The shadow trader

    A backtest said my arbitrage strategy was profitable. A shadow trader running inside the production process, against the same order book, said it was off by 65 basis points. The architectural difference between the two was the part the backtest ignored. That part decided the outcome.

    ·12 min read

  3. The Two Webs

    Documents and applications need different runtimes. Quietly, that separation is already happening, and the architects of the modern web are starting to admit it.

    ·18 min read
C

Cloud

Platforms · infrastructure · scale

1 essay

  1. My homelab is not for reliability

    My homelab has worse uptime than the €17 server that runs my actual product. I keep it anyway, because reliability was never the point. It is the zero-stakes place where I rehearse the operational disciplines that matter where it counts.

    ·13 min read
D

AI

Models, tooling, and the working day

2 essays

  1. Auditing my own AI workflow

    I had a seventeen-turn conversation with another instance of Claude about how I use Claude Code. The valuation it gave me was wrong in both directions: inflated where it guessed, deflated where I had quietly underdescribed what I had built. The interesting result was not the score. It was the bias on both sides.

    ·17 min read

  2. The most useful bug I shipped

    I ported my Python crypto bot to Rust in roughly ten hours with AI peer review on every fix. I got a 138× speedup. The speedup was generating 244 arbitrage opportunities that did not exist. The fix taught me more about AI-assisted development than the port itself.

    ·13 min read
E

Career

Tech in Italy · sponsorship · pivot

1 essay

  1. Three bugs before the market

    I built a triangular arbitrage bot, ran it in Tokyo with 21ms latency, and got zero opportunities. I almost concluded the market was efficient. Then I asked the question I should have asked first, and found three bugs in my own code.

    ·11 min read