{"canonicalUrl":"https://humetric.net/p/profile-8d531e2f346d4c739c48a2d039b8557e","profileVersion":"f8882523-f070-43ff-b744-a4b51eeb7306","lastUpdated":"2026-10-19T14:45:38.225027+00:00","profile":{"display_name":"TZU KUAN YU","last_updated":"2026-10-19","slug":"profile-5450a8005bff460a","summary":"A computer science graduate student with an interdisciplinary background and strong self-learning capabilities, researching Graph Neural Networks for financial fraud detection. Proficient in C/C++ and Python, with hands-on experience in embedded operating systems, kernel-level scheduling, resource management, emerging storage architectures, and system design.","skills":["C","C++","Ceiling Priority Protocol","Earliest Deadline First Scheduling","Embedded System","Embedded Systems","Git","GitHub","Graph Neural Networks","Linux","Non-Preemptive Critical Sections","Python","PyTorch","Rate Monotonic Scheduling","Real-Time Operating Systems","RTOS","μC/OS-II"],"educations":[{"id":"b75130b0-5b5f-4f78-bfd4-576262a68d26","person_id":"60f8f649-fd82-4e99-bbb3-0fc5d473d6cb","school":"National Taiwan University of Science and Technology","degree":"Master","field":"Computer Science and Information Engineering","start_year":2025,"end_year":2027},{"id":"d5fa190e-d52b-41c0-afb0-e[contact removed]","person_id":"60f8f649-fd82-4e99-bbb3-0fc5d473d6cb","school":"National Central University","degree":"Bachelor","field":"Atmospheric Science","start_year":2021,"end_year":1960}],"projects":[{"id":"e9024fa5-8e98-41d9-95bb-fe70b115934f","person_id":"60f8f649-fd82-4e99-bbb3-0fc5d473d6cb","name":"RTOS Kernel-Level Scheduling and Resource Management Development","description":"Implemented Rate Monotonic (RM) and Earliest-Deadline-First (EDF) dynamic scheduling algorithms at the kernel level, successfully managing preemption and deadline logic for multiple periodic tasks. Addressed multi-threading race conditions by manually implementing Non-Preemptive Critical Section (NPCS) and Ceiling Priority Protocol (CPP), effectively resolving priority inversion issues common in embedded systems. Directly manipulated Task Control Block (TCB) linked lists to build a custom context switch recorder, accurately tracking task switching behaviors and system latency. The work involved μC/OS-II kernel-level implementation.","url":null,"technologies":["C","μC/OS-II","Real-Time Operating Systems","Rate Monotonic Scheduling","Earliest Deadline First Scheduling","Non-Preemptive Critical Sections","Ceiling Priority Protocol"],"position":0},{"id":"68fcfc1b-3fce-4046-bf0c-4f2c2552dfbe","person_id":"60f8f649-fd82-4e99-bbb3-0fc5d473d6cb","name":"Insurance Fraud Detection based on Graph Neural Networks (GNN)","description":"Modeled complex relationships between insurance agents and policyholders to research and extract feature representations of anomalous behaviors. Built Graph Neural Network models using Python to process large-scale graph data, enhancing the robustness of the anomaly detection model.","url":null,"technologies":["Python","Graph Neural Networks","PyTorch"],"position":1}]},"chunksVersion":"profile-chunks-v4","approvedDiscoveryQueries":[],"evidence":[{"id":"identity:0","type":"identity","ordinal":0,"locale":"en","text":"TZU KUAN YU","fieldKeys":["display_name"]},{"id":"summary:0","type":"summary","ordinal":0,"locale":"en","text":"TZU KUAN YU. Summary / 簡介: A computer science graduate student with an interdisciplinary background and strong self-learning capabilities, researching Graph Neural Networks for financial fraud detection. Proficient in C/C++ and Python, with hands-on experience in embedded operating systems, kernel-level scheduling, resource management, emerging storage architectures, and system design.","fieldKeys":["summary"]},{"id":"skills:0","type":"skills","ordinal":0,"locale":"en","text":"TZU KUAN YU. Skills / 技能: C, C++, Ceiling Priority Protocol, Earliest Deadline First Scheduling, Embedded System, Embedded Systems, Git, GitHub, Graph Neural Networks, Linux, Non-Preemptive Critical Sections, Python, PyTorch, Rate Monotonic Scheduling, Real-Time Operating Systems, RTOS, μC/OS-II","fieldKeys":["skills"]},{"id":"discovery:0","type":"discovery","ordinal":0,"locale":"en","text":"TZU KUAN YU. Expertise / 專長: C, C++, Ceiling Priority Protocol, Earliest Deadline First Scheduling, Embedded System, Embedded Systems, Git, GitHub, Graph Neural Networks, Linux, Non-Preemptive Critical Sections, Python","fieldKeys":["display_name","skills"]},{"id":"education:0","type":"education","ordinal":0,"locale":"en","text":"TZU KUAN YU. Education / 學歷: Master. Computer Science and Information Engineering. National Taiwan University of Science and Technology","fieldKeys":["educations"]},{"id":"education:1","type":"education","ordinal":1,"locale":"en","text":"TZU KUAN YU. Education / 學歷: Bachelor. Atmospheric Science. National Central University","fieldKeys":["educations"]},{"id":"project:0","type":"project","ordinal":0,"locale":"en","text":"TZU KUAN YU. Project / 專案: RTOS Kernel-Level Scheduling and Resource Management Development. Built with / 使用技術: C, μC/OS-II, Real-Time Operating Systems, Rate Monotonic Scheduling, Earliest Deadline First Scheduling, Non-Preemptive Critical Sections, Ceiling Priority Protocol. Implemented Rate Monotonic (RM) and Earliest-Deadline-First (EDF) dynamic scheduling algorithms at the kernel level, successfully managing preemption and deadline logic for multiple periodic tasks. Addressed multi-threading race conditions by manually implementing Non-Preemptive Critical Section (NPCS) and Ceiling Priority Protocol (CPP), effectively resolving priority inversion issues common in embedded systems. Directly manipulated Task Control Block (TCB) linked lists to build a custom context switch recorder, accurately tracking task switching behaviors and system latency. The work involved μC/OS-II kernel-level implementation.","fieldKeys":["projects"]},{"id":"project:1","type":"project","ordinal":1,"locale":"en","text":"TZU KUAN YU. Project / 專案: Insurance Fraud Detection based on Graph Neural Networks (GNN). Built with / 使用技術: Python, Graph Neural Networks, PyTorch. Modeled complex relationships between insurance agents and policyholders to research and extract feature representations of anomalous behaviors. Built Graph Neural Network models using Python to process large-scale graph data, enhancing the robustness of the anomaly detection model.","fieldKeys":["projects"]}]}