HSF student pages for GSoC 2025
Enhancing Keras Parser and JAX/FLAX Integration
Prasanna Kasar
Developed a parser within SOFIE to parse Machine Learning models trained with Keras. Rewrote the existing parser in Python, which was previously written in C++. Added support for parsing missing layers, such as Pooling and LayerNormalization, and wrote unit tests for the parser.
Intelligent Logging Pipeline
Osama Ahmed Tahir
An intelligent logging pipeline for NopayloadDB that integrates log aggregation, anomaly detection, and monitoring. It improves reliability and maintainability in large-scale HEP experiments It is deployed on Minikube cluster while using a deep learning model for real-time anomaly detection.
Neural (De)compression for High Energy Physics
Yolanne Lee
In high-energy physics experiments such as those at CERN’s ATLAS project, immense volumes of data are generated. This project explores the feasibility for “precision upsampling” using deep generative models to be used to reconstruct high-precision floating-point data from aggressively compressed representations.
Full list of projects in 2025
- Abhishikth Mallampalli - Highly Granular Quantization for CICADA - Lino Gerlach, Jennifer Ngadiuba
- Andrei Girjoaba - Integrating Support for Google XLS in hls4ml - Vladimir L, Dimitrios Danopoulos
- Carter Capetz - Physics-Constrained Autoencoders: Intelligent Compression in High Energy Physics - James Smith
- JohnKala - Background Enrichment augmented Anomaly Detection (BEAD)– Contrastive VAE & Transformer Architecture - Pratik Jawahar, Sukanya Sinha
- Kriti Mahajan - RNTuple in JSROOT - Sergey Linev, Giacomo Parolini
- MytsV - Rucio WebUI Revamp - Martin B, Mayank Sharma
- Osama Tahir - Intelligent Log Analysis for the HSF Conditions Database - Ruslan Mashinistov, Michel Villanueva, John S. De Stefano Jr.
- Prasanna Kasar - TMVA SOFIE - Enhancing Keras Parser and JAX/FLAX Integration - Lorenzo Moneta, Sanjiban Sengupta
- S. Akash - TMVA SOFIE - GPU Support for Machine Learning Inference - Lorenzo Moneta, Sanjiban Sengupta
- Tarun Nandi - Data Representation Optimisation for Generative Model-based Fast Calorimeter Shower Simulation - Piyush Raikwar, Peter McKeown
- Yolanne Lee - Neural (De)compression for High Energy Physics - Maciej Szymański, Peter Van Gemmeren
- Jason Wu - Evaluating CVMFS for Machine Learning Model Distribution - Lorenzo Moneta, Valentin Volkl
- Petro Zarytski - Improve automatic differentiation of object-oriented paradigms using Clad - Vassil Vassilev, David Lange
- Karan Singh - Benchmarking Sustainability of Classical & Quantum Algorithms for particle trajectory reconstruction - MiriamLucioMartinez, Arantza Oyanguren
- Kossi Glokpor - Implementing a deprecation system for Ganga - Alexander Richards, Ulrik
- Madlani Shivam - Extending support on custom kernels with virtme-ng - Valentin Volkl, Georgios Christodoulis
- Salvador de la Torre Gonzalez - CARTopiaX: an Agent-Based Simulation of CAR T-Cell Therapy built on BioDynaMo - Vassil Vassilev, Lukas Breitwieser
- Abhinav Kumar - Implementing Debugging Support for xeus-cpp - Vassil Vassilev, Vipul Cariappa
- Abdelrhman Elrawy - Support usage of Thrust API in Clad - Vassil Vassilev, Alexander Penev
- Aditi Milind Joshi - Implement and improve an efficient, layered tape with prefetching capabilities - Vassil Vassilev, David Lange
- Aditya Pandey - Using ROOT in the field of genome sequencing - Vassil Vassilev, Martin Vassilev
- Jiayang Li - Enable automatic differentiation of OpenMP programs with Clad - Vassil Vassilev, Martin Vassilev
- Maksym Andriichuk - Implement activity analysis for reverse-mode differentiation of (CUDA) GPU kernels - Vassil Vassilev, David Lange
- Rohan Timmaraju - Enhancing LLM Training Efficiency with Clad for Automatic Differentiation in C++ - Vassil Vassilev, David Lange
- Sakshi Kumar - GreenML@CERN – A Comprehensive Framework for Energy-Efficient Scientific Machine Learning - Caterina Doglioni