No publications yet, it is still too early in my PhD!
Sept. 2021: Systèmes numériques at ENS Paris
I gave a short talk at Devfest Toulouse 2019, explaining how it was humanly possible to write useful Brainfuck programs. After explaining the syntax of the language, I explained some useful code patterns such as conditions or array management, which allow to write complex programs.
I gave a talk that was aimed to explain a computer science notion for high-school student. I decided to explain the max-flow min-cut theorem and the Ford-Fulkerson algorithm, applied on a problem of deciding whether a baseball team can still qualify or not. I definitely lacked a lot of presentation skills (you can count the “euuuuh”), however I think I found the simplest explanations for the max-flow alogrithm.
I worked five months in the Prosecco team at INRIA. I wrote a global and local specifications for the online phase of the SPDZ2k protocol, and proved that they computed the same thing.
I worked five months in the CakeML team to implement a verified linear-scan register allocator. The motivation was that the register allocation was the slowest part of the compilation process, hence a faster algorithm could drastically reduce compilation time.
I worked two months on cache timing attacks (before Spectre or Meltdown were known). I found a procedure to analyse a cryptographic library and automatically find parts of the code which break secret independance, and might leak a private key to an attacker running on the same computer.