CakeResume Talent Search

Work Experience Egis Techno logy I nc. Achievements Mass production of LG mobiles (V60 ThinQ, G8X ThinQ, Velvet, Wing) Alipay certification EMVCo SBMP security evaluation certification An image counter written in Python for data collection and simulation test on Android devices Replacing manually operation Reducing time cost of data classification from 10 minutes to 0 in one collection Decreasing human errors Making it configurable to be applied on multiple projects Positions Staff Software Engineer OctNow Software Engineer MayOct 2022 Responsibilities Production support Integrating fingerprint SDK, including client application (CA) and trusted application (TA), into mobiles/tablets. Customizing

inventory, risk assessment, management review meeting tracking and control 1-3. Internal and external audit 1-4. Business Continuity Plan (BCP) management, social engineering, information security health diagnosis, vulnerability scanning and penetration testing management examination tracking 2. System protection and control review 2-1. New system classification, online control item review, risk assessment 2-2. Maturity assessment of information security governance in technical service center 2-3. Writing and revising the information security maintenance plan of the whole center 2-4. Signing of operation level agreement (OLA) and service level agreement (SLAService catalog

預測結果，預測結果可以選擇是單純的判斷是否為惡意軟體，或者是惡意家族分類，最後做出的成果準確率達到99%。 Few-shot Malware Classification 透過cuckoo sandbox蒐集出的資料，提取呼叫的API call序列，將序列透過單位時間內執行的指令數，將其轉換成圖片，並結合深度學習

辨識與分類 產學計畫 碩士論文 施孟宏 Hiro New Graduate Xizhi, New Taipei During the studies, Python and deep learning were used to solve the projects: * Silver medal in Kaggle competition: APTOS 2019 Blindness Detection * Sphygmomanometer digits recognition system * Radar signal classification and deinterleaving Experience with LiDAR and Jetson edge computing devices Each project is the result of teamwork For more project descriptions, please see my portfolio website https://hirortw101.github.io/ [email protected] Python Deep Learning PyTorch FastAI Keras ultralytics/yolov5

式能力檢定(CPE) 3題 Dec 2019 Education 國立台灣科技大學 National Taiwan University of Science and Technology 資訊工程 • 指導教授 : 李漢銘/鄭欣明 教授研究領域 : Malware Detection and Classification 國立台灣科技大學 National Taiwan University of Science and Technology 資訊工程 •畢業專題:單視角及多重視角影像人體辨識及計數的應用探討 Project 單視角

Transformer to classify news documents by PyTorch, Top 2 accuracy in class Implement VAE and GAN to create anime face by pyTorch Build DQN and Actor-Critc RL model on Lunarlander using PyTorch Machine Learning EE, NYCU (交大電子所) & MTK Spring 2022 , GRADE: A Train an image classification model on CIFAR-100 with limited size, 100MB Build ResNet-50 and use SAM Optimizer (2021, Google) Achieve almost 83% accuracy Final project is building road image segmentaion model on mediaTek chip Using U-Net and post-training quantization to compress tflite model AchievemIOU, Top 3 in class Data

Bioinformatics Algorithms , Biomedical Data Mining, Computer Organization, Digital Systems Design, Digital Signal processor Thesis title：Using deep learning to predict T-cell receptors binding specificity Project Experience C++ Greedy snake design Verilog Gobang in FPGA Single Cycle implementation Python Data structure algorithm implementation Image processing implementation Pytorch Image Classification ( CNN, Semi-Supervised learning, S elf-Supervised learning, VIT, ACGAN ) Sequence to Sequence( Transformer, BERT ) Semantic segmentation ( FCN, U-net) Face and Digit Generation ( SC-GAN) Lunar Lander Game ( Reinforcement Learning ) Skull fracture detection ( Faster R-CNN ) Django Web backend API Design Costco notice of goods Linux Simple shell