arielb57

I am a self-taught developer in Paris. Most of this year went into two things: the backend of VYVE, a marketplace, and forge, a pipeline that reads the week's engineering news and turns it into small tools that actually get finished. Nearly everything public on this account came out of the second one. Most of it is about money or markets, because that is where a cent off is a bug and “close enough” does not survive an audit.

The habit I care about most is that a tool should say what it knows. When the exact answer is reachable, I get it, in integers or rationals rather than floats, and the README says so. When it is not, the best answer goes on the screen next to a bound that is actually proven, so the worst case is in the output rather than in someone's head. When the real finding is that the usual implementation is wrong, the counterexample ships with the fix. I write the Limitations section of each README first. Someone will find them anyway.

One example of why I bother. The README of netmin used to say the exact solver took 0.1 s at 24 parties. My benchmark stopped at 20. The number for 24 had never been measured; I had extrapolated it. When I finally ran it, it was 151 ms on average and 240 ms at worst. The correction is in the commit history and the README now says what the benchmark prints. That is the rule I try to hold everything to: if there is a number, there is a script that produces it. It is also why every figure on this page is read from the GitHub API when the page is built, not typed in.

VYVE is private, so it is not in the list above. It is Postgres with row-level security everywhere, an escrow payment flow, and the legal and tax structure that has to match it: 343 commits and 273 test files since May 2026. The hard part was not the code. It was deciding who is allowed to see what, and then proving the database enforces it rather than hoping the application layer does.

forge reads seven engineering feeds every morning, ranks what it finds by how many sources agree, turns the survivors into specifications, builds them, and then does its best to reject its own output. The quality gate counts assertions and checks how many of them compare a literal to itself; too many and the build fails. It never publishes on its own. Every run ends in a review queue where I read the result. The repositories it produced carry the forge-built topic, and the cards above say so.

How this was made. Everything here was written with AI assistance. I would rather say that than have someone work it out. The tool repositories came out of forge from a specification and were reviewed by me before publication; forge itself, this site and VYVE I built directly, deciding and correcting as I went. Either way I read all of it before it goes public. If I cannot explain a repository, it should not be on here.

I apply to university next year. The time until then goes on things that are worth showing rather than things that are easy to count. Every number on this page is read from the GitHub API once a day; if one looks small, it is because it is small.