Projects

This page showcases my research software projects, implementation work, and contributions to the machine learning and optimization communities.

Project Categories

Research Software Projects

My projects focus on developing efficient implementations of machine learning algorithms, optimization methods, and research infrastructure tools.

� 6DIMCOCO - Cross-Modal Contrastive Learning

6DIMCOCO A framework for six-dimensional contrastive learning that enables cross-modal understanding across different data types. This project implements novel contrastive learning approaches for multi-modal representation learning.

Key Features:

  • Multi-modal contrastive learning framework

  • Efficient implementation for large-scale datasets

  • Comprehensive evaluation suite

  • Extensible architecture for different modalities

�️ JERICHO - Adversarial Robustness Research

JERICHO A research environment for studying adversarial robustness in machine learning models. JERICHO provides tools for analyzing model vulnerabilities and developing more robust training methods.

Key Features:

  • Adversarial attack and defense frameworks

  • Comprehensive robustness evaluation metrics

  • Integration with popular ML libraries

  • Reproducible experimental protocols

� ML-SLURM-Template - HPC for Machine Learning

ML-SLURM-Template A comprehensive template for running machine learning experiments on SLURM-based high-performance computing clusters. This project streamlines the process of scaling ML research to large computational resources.

Key Features:

  • Ready-to-use SLURM job templates

  • Distributed training configurations

  • Resource management best practices

  • Automated experiment tracking

📊 PGDVisualisation - Optimization Visualization

PGDVisualisation Visualization tools for projected gradient descent methods, particularly useful for understanding adversarial example generation and optimization landscapes in machine learning.

Key Features:

  • Interactive optimization trajectory visualization

  • Support for various PGD variants

  • Analysis tools for optimization behavior

  • Educational and research-focused interfaces

Development Philosophy

Research-Driven Development

All projects emerge from active research needs and are designed to advance the state of understanding in machine learning and optimization.

Open Science Commitment

  • All code is open source and well-documented

  • Reproducible research practices

  • Community engagement and collaboration

  • Long-term maintenance and support

Technical Excellence

  • Efficient implementations optimized for performance

  • Comprehensive testing and validation

  • Clear documentation and examples

  • Modular, extensible architectures

For detailed information about specific research software, see Research Software. Visit my GitHub profile for the complete codebase.