Identified a gap in the team's existing flow and developed an unsupervised algorithmic solution. Successfully implemented this solution, demonstrating 30% decrease in false positive rate.
Modernized algorithmic infrastructure to support parallelism, easy configurability and flexible interfacing.
Implemented automatic hyperparameter tuning with Optuna, resulting in 15% increase in main AUC metric and significantly reducing experimental workload, saving the team countless hours.
Mentored junior developers, providing guidance on best practices in algorithm development and optimization techniques.
MSc. of Particle Physics
Tel Aviv University
01.2021 - Current
Applied Computer Vision / Deep Learning thesis – Researching, training and tuning deep learning bases solutions
Goal - Energy spectrum and position reconstruction of a particle detector (E-Calorimeter) “images” for future LUXE experiment.
Software/Algorithm developer
Kama Research
01.2020 - 01.2021
Design and maintain an infrastructure for a multi-service software (python), used for online data analysis and intravascular visualization
Develop supporting machine learning algorithms for mapping the heart during cardiac catheterization surgeries.
Algorithm developer
Navin
01.2019 - 01.2020
Develop explicit algorithms for indoors mapping and navigation purposes
Develop interactive simulation and visualization tools using JS and HTML.