Eman Maali

I joined the Computational Privacy Group at Imperial College London in June 2025 as a Research Fellow, working with Yves-Alexandre de Montjoye and Matthew Wicker. My research focuses on the robustness of machine learning systems, developing evaluation methods and formal frameworks that predict how models and multi-agent AI pipelines fail when deployed in adversarial or real-world conditions. I study this across two settings: how ML models used for identification and classification tasks degrade when the environment shifts, and how multi-agent AI systems can be compromised at the pipeline level even when each component appears robust. My current work covers privacy attacks against ML models and data releases, LLM fingerprinting, and adversarial robustness of multi-agent AI pipelines.

I received my PhD from the Department of Computing at Imperial College London in May 2025, where I was a member of the Adaptive Emergent Systems Engineering Laboratory (AESE) under the supervision of Professor Julie McCann. My doctoral research evaluated the robustness of ML-based IoT device identification systems under real deployment conditions, including device mode changes, spatial and temporal distribution shift, and sampled network traffic.

Before my PhD, I completed an MSc in Electromagnetic Sensor Networks at the University of Birmingham (2017) and a BSc in Computer Systems Engineering at Birzeit University, Palestine (2013), and taught as a lecturer at Birzeit University from 2017 to 2019.