The rigor of critical software, applied to the web

I am passionate about software design. A passion I owe as much to my education, a Master's in Software Fundamentals at Université Paris Diderot Paris 7, as to my many experiences.

As part of strategic assignments within the French Ministry of Finance and the Ministry of the Interior, I led critical projects — where the slightest mistake comes at a high price.

I then created Carto-SI, a software that maps the information systems of large corporations. Over more than ten years, this solution allowed me to support large organizations where the information system is highly strategic and sensitive — and it even took me to Toronto, working with the National Bank of Canada.

Back in France, one demand followed me through every assignment: software quality — the certainty that a simple change, sometimes a single line of code, will not quietly break another part. No tool answered it simply. I thought the problem was simple; digging deeper, I measured its true algorithmic difficulty and understood why no existing solution truly addressed it. It was by pushing this reflection further that I forged a relevant, robust and deterministic algorithm — Delta-QA — with the ambition of making it the best visual regression solution on the market.

This software is deployed at our clients', in strict respect of their security: no data ever leaves their information system. The entire design was guided by one demanding constraint — the fewest possible false positives and false negatives — while flagging the differences a human eye would notice.

It is long and difficult work, made possible today by all the experience I have accumulated and by the workforce of a team of AI agents I orchestrate — from analysis to marketing — to explore, test and harden every building block of the product.

A quality tool must be reliable, reproducible and explainable — never a black box.