A team project focused on training and testing a number of classification
models that predict where a crash would occur, as well as predict the severity
of the crash.
Implementation of a system simulating loop extrusion using pure Python and
dynamics simulating thermal noise (as a Python layer over OpenMM), accelerated by GPU (CUDA).
Different deep learning architectures implemented in multiple different common libraries/frameworks for model building
and testing.
Data-processing pipelines for common bioinformatics NGS (Next-Generation Sequencing) applications, written in WDL
using Docker containers to ensure reproducability, version control, and ease of use.
Full-stack webapp with database (PostgreSQL), backend API (Flask) and dedicated frontend (ReactJS). Displays Zillow real estate data
as dynamic plots using ChartJS.