Jun 2025 – Jul 2025 · 1 mo · Delhi, India · On-site

• Conducted an in-depth literature review of adversarial-ML attacks (FGSM, PGD, CW), classifying techniques into gradient, optimization and boundary-based families.
• Presented key takeaways to the research mentor, outlining experiment ideas to evaluate neural-network robustness under real-world attack scenarios.