- Engineered custom robotic platforms and MuJoCo simulation environments to support reinforcement learning research and systematic sim-to-real transfer studies - Implemented and trained state-of-the-art deep RL algorithms in TorchRL, with emphasis on sample efficiency, reward convergence, and policy stability - Designed reproducible experimental protocols to quantify sim-to- real generalization gaps and evaluate policy robustness under domain discrepancies - Delivered technical workshops to disseminate best practices in reinforcement learning experimentation and robotics-driven AI systems

- Conducted penetration testing on web applications and corporate infrastructure following OWASP methodology, performing reconnaissance, exploitation, privilege escalation, and lateral movement assessments - Identified critical vulnerabilities including injection flaws, misconfigurations, and access control weaknesses using industry-standard offensive tooling, delivering risk-prioritized remediation reports - Implemented enterprise email security hardening measures to mitigate phishing, spoofing, and account compromise risks through protocol enforcement and secure configuration practices

Designed and implemented a human-machine interface for a secure OS configuration tool, featuring a local web interface built with Python Flask for streamlined management and configuration