Mark Santolucito

I am a PhD student in the Dept. of Computer Science at Yale in the C2 (Creative Consilience) program. My advisor is Ruzica Piskac, and my research focus is Program Synthesis and Computer Music.


Misconfiguration errors account for the majority of downtime in real-world deployed systems. ConfigV is a machine learning based system to automatically learn specifications for configuration files and check for correctness. It uses an association rule learning approach to build sets of predicates any correct configuration files must satisfy.

ConfigV is written in Haskell, and an open source implementation is available inside a virtual machine image. Simply load the image into a new virtaul machine and get started. The username and password are both 'cavae'. Once you are in, to see the program in action, execute the following commands. More settings can be found in the 'src/Settings.hs' file.

    cd configErrorDetect/code2
    cabal run