Sungjae Lee (James)
Hi! I am a first-year PhD student in the NOW Lab at The Ohio State University in the Department of Computer Science and Engineering. My research goal is to build efficient parallel systems for AI workloads. To achieve this goal, I conducted research projects on Deep Learning Systems, Serverless Computing, Edge Computing, and Spot Instances. Before joining OSU, I earned my Master’s and Bachelor’s degree at Kookmin University and worked as an Undergraduate/Graduate Research Assistant in the DDPS Lab. At Kookmin University, I focused on a project about cost and time-efficient deep learning systems with my advisor, Professor Kyungyong Lee.
Education
[Doctoral Degree] The Ohio State University, Department of Computer Science and Engineering (ongoing)
[Master’s Degree] Kookmin University, Department of Computer Science
[Bachelor’s Degree] Kookmin University
- Major in Business Administration
- Minor in Computer Science
Research Experience
Graduate Research/Teaching Assistant
- NOW Lab, The Ohio State University
Graduate Research Assistant
- DDPS Lab, Kookmin University
- Analysis of Spot Instances for Cost-effective System (IISWC 2022)
- Performance Modeling of Deep Learning Workloads (IEEE Big Data 2022)
Undergraduate Research Assistant
- DDPS Lab, Kookmin University
- Serverless Inference System for Ensemble Models (KSC 2021, Best Presentation Paper Award)
- Cloud Face Recognition with Deep Learning based Edge Computing (KSC 2019)
- Dept. of Statistics, UC Irvine
Teaching Experience
Teaching Assistant
- Cloud Architect Course, Samsung
- Scalable Deep Lerning Inference System with Serverless Computing
- Deep Learning Institute, NVIDIA
- Fundamental of Accelerated Computing with CUDA C/C++
- Fundamental of Deep Learning for Computer Vision
- Kookmin University
- Python Programming (teaching)
- File Processing (grading)
- Database (grading)
Publication
International Publication
- Sungjae Lee, Jaeil Hwang, and Kyungyong Lee. “SpotLake: Diverse Spot Instance Dataset Archive Service.” 2022 IEEE International Symposium on Workload Characterization (IISWC) [Paper] [Slide] [Demo]
- Sungjae Lee, Yoonseo Hur, Subin Park, and Kyungyong Lee. “PROFET: Profiling-based CNN Training Latency Prophet for GPU Cloud Instances.” 2022 IEEE International Conference on Big Data [Paper] [Slide] [Demo]
- Sungjae Lee*, Yeonji Lee*, Junho Kim, and Kyungyong Lee. “RnR: Extraction of Visual Attributes from Large-Scale Fashion Dataset”, Workshop on Large Media Dataset at 2019 IEEE International Conference on Big Data (*equal contribution) [Paper]
Domestic Publication (Korean)
- Sungjae Lee, Jaeghang Choi, and Kyungyong Lee. “Serverless Inference System for Ensemble Models”, Best Presentation Award at KSC 2021 [Paper]
- Junsang Park* and Sungjae Lee*. “Classification of High-risk Suicide Group’s SNS Post based on Bi-LSTM”, KCC 2021 (*equal contribution) [Paper]
- Sungjae Lee, Jaeghang Choi, Unho Choi, and Kyungyong Lee. “Scalable Recommender System based on Serverless Computing”, KSC 2020 [Paper]
- Sungjae Lee, Dongwook Kim, Hyeonktae Choi, Subin Park, and Kyungyong Lee. “Accelerating of Cloud Face Recognition with Deep Learning based Edge Computing”, KSC 2019 [Paper]
- Sungjae Lee, Hyeonwoong Woo, Hyeokman Kim, and Kyungyong Lee. “Detect Posts which is using Hashtag Generator with Spectral Clustering”, Best Paper Award at KCC 2019 [Paper]
Personal Project
Flex Ads, 2019
- Personalized Digital Signage with Efficient Cloud-Edge Face Recongition
AAA, 2018
- Predict States of Abandoned Animals with Machine Learning Algorithms
Fumby, 2016
- Connect 3D-Printed Robot Hand to MNIST Application