OptiAttack

OptiAttack is a professional-grade tool designed to test and improve the robustness of AI models by generating adversarial attacks. Adversarial attacks are subtle perturbations to input data that cause machine learning models to make incorrect predictions. OptiAttack automates the generation of these attacks, helping security researchers and ML engineers identify vulnerabilities in their models. This project was developed during my internship at Erciyes University in Turkey, requiring strong English communication skills and international collaboration. The tool features an intuitive React dashboard for configuring attacks and visualizing results, backed by powerful Python algorithms. Working with a diverse, international team taught me valuable lessons in cross-cultural communication, distributed development, and professional software engineering practices.

• Developed interactive React dashboard for attack configuration
• Modified Python command-line tools for better usability
• Performed comprehensive end-to-end testing
• Collaborated with international team in English
• Analyzed and documented attack results in JSON format