Mingjing Xu is an embedded-AI engineer turned machine-learning scholar, whose research bridges mathematical optimization and trustworthy, edge-deployable AI. He is first author of PSMGD, an AAAI-25 paper that accelerates multi-objective learning with periodic gradient scheduling; he also exposed stealthy packet-level exploits on WiFi sensing in an ACM MobiCom-24 study, and co-developed a magnetically guided origami robot for targeted therapy, presented at ICRA-24. Before academia he built real-time firmware for IoT radio units at Jabil Circuit, and he has taught Android/Kotlin app development at Temple University. Mingjing now focuses on secure federated and multi-task learning, combining rigorous theory with systems implementation to make AI efficient, robust, and socially responsible.