██▀███  ▓██   ██▓ ██▒   █▓▓█████ ▒██   ██▒ ▄▄▄       ███▄ ▄███▓
▓██ ▒ ██▒▒██  ██▒▓██░   █▒▓█   ▀ ▒▒ █ █ ▒░▒████▄    ▓██▒▀█▀ ██▒
▓██ ░▄█ ▒ ▒██ ██░ ▓██  █▒░▒███   ░░  █   ░▒██  ▀█▄  ▓██    ▓██░
▒██▀▀█▄   ░ ▐██▓░  ▒██ █░░▒▓█  ▄  ░ █ █ ▒ ░██▄▄▄▄██ ▒██    ▒██
░██▓ ▒██▒ ░ ██▒▓░   ▒▀█░  ░▒████▒▒██▒ ▒██▒ ▓█   ▓██▒▒██▒   ░██▒

I'm a software engineer focused on DevSecOps, automation and agentic development.

I currently work on cybersecurity and automation inside a large CAC 40 industrial group.

Most of my personal experiments revolve around one idea:

The agent should know the codebase. I should know what the product is supposed to do.

I want AI to understand repositories, implement from specifications, test its work and be challenged by other agents and automated security checks before anything gets merged.

I'm intentionally comfortable across the whole stack because I think modern development is becoming less about knowing every line of code and more about understanding architecture, boundaries, data flows and core concepts.

The agent can learn the codebase from its documentation and repository context. My job is to understand the system well enough to specify what it should become and verify that it gets there safely.

If it's open source and cool, there's basically a 0% chance I haven't forked it already.

Usually it goes something like:

find cool project
      ↓
fork it
      ↓
"just one small change"
      ↓
three features later
      ↓
why am I maintaining this now?

I strongly believe every 30-minute manual task deserves at least three hours of automation.

And more seriously:

If I'm going to spend a huge part of my life building things, I'd rather build things I think are worth it.