Summary
Krishna Nakka is a privacy-preserving ML researcher based in Munich with eight years of experience bridging cutting-edge academic research and applied industry work. Currently at Huawei Munich Research Center, he investigates PII extraction and model internals of LLMs, authoring multiple workshop and arXiv papers on jailbreaking, activation steering, and private-data leakage. He holds a PhD from EPFL where his work on interpretable architectures, adversarial attacks and transferable attacks earned top-tier conference acceptances (ECCV, NeurIPS, ICCV) and a postdoc focused on simultaneous detection and tracking for sports analytics. Earlier roles at Samsung and internships at leading labs gave him deep hands-on skills in object detection, occlusion handling, and resource-efficient models for edge devices. Krishna combines strong theoretical rigor with practical threat-driven experiments—an uncommon blend that helps translate privacy risks into concrete mitigation strategies. He is comfortable working across federated settings, model interpretation, and real-world deployment constraints, often exploring reward-based and obfuscation techniques to balance utility and privacy.
8 years of coding experience
2 years of employment as a software developer
MPC, MPC at Narayana Junior College, Nallakunta (NGIC)
Doctor of Philosophy - PhD, Computer Science, 5.6, Doctor of Philosophy - PhD, Computer Science, 5.6 at EPFL (École polytechnique fédérale de Lausanne)
Bachelor of Technology (BTech) and Master of Technology (MTech), Signal Processing, Electrical Engineering, 8.9/10, Bachelor of Technology (BTech) and Master of Technology (MTech), Signal Processing, Electrical Engineering, 8.9/10 at Indian Institute of Technology, Kharagpur