fridex

Participated in the initial database design, and implemented core parts of the SBOM Hub(an internal service for SBOM aggregation. Participated in TUF, in-toto, SLSA, GUAC upstream meetings, OpenSSF community meetings and applied related practices internally. Tested software related to Datadog’s Python distributions that use TUF. Designed and implemented a new Windows Code Signing process that uses AWS KMS. Authored tools such as pipctl, Yorkshire. Helped to evaluate third-party vendors and their suitability for Datadog products.

Implementation of the Thoth cloud-based Python resolver based on reinforcement learning, author of micropipenv. Interviewed, guided, and led new engineers and interns; provided technical guidance on theses for University students. Led internal process for publishing and releasing Python packages on a company-wide Python package index for internal and external use; co-authored or authored 16 patent proposals; five of which are active patents, three are pending, designed and implemented most of Dependency Monkey and Thoth backend. Presented work at conferences including NeurIPS 2020 or PyCon US. Collaborated with IBM Research, participated in Google’s TensorFlow SIG build, and worked within Python upstream communities.

Implementation of C/C++ LLVM module in RetDec, focused on reconstruction of code blocks that are optimized by compilers (instruction idioms, optimization of standard C/C++ library functions) on different architectures (Intel, MIPS, ARM). The module outperformed the industry standard HexRays decompiler in reconstruction of optimized code. The work is still used in the open-source version of RetDec. Published related research papers in ComSIS and IEEE magazines. The module helped with malware analyses.