Woohyeon Baek

Perspectives on Computer Systems

• Runtime (dynamic) & Compiler (static) > Hardware (dynamic)
• Correctness > Performance > Energy-efficiency
• General-purpose > Domain-specific
• Systems > Applications
• Deterministic > Non-deterministic > Probabilistic

Research Interests

• High-performance computing
• Runtimes and Compilers
• Distributed & Parallel Systems
• Disaggregated Memory & Heterogeneous Systems

Research Experience

• Visiting Researcher, Pacific Northwest National Laboratory, 09/2026 - Present
• Ph.D. Intern, Pacific Northwest National Laboratory, 07/2025 - 12/2025
  • Developing runtime and compiler support for exascale computing systems

Education

• Ph.D. in Computer Science and Engineering, The Pennsylvania State University, 08/2024 - Present
  • Advisor: Prof. Mahmut Taylan Kandemir
• B.S. in Computer Science and Engineering, Seoul National University, 03/2017 - 02/2023
  • Thesis: “Liquid: Mix-and-Match Multiple Image Formats to Balance DNN Training Pipeline”

Publications

• 2 papers currently under review
• Myungjun Son, Woohyeon Baek, Mehrdad Mahdavi, Mahmut Kandemir. “SpotFed: Cost-Efficient Federated Learning with Spot Instance Compensation”. CCGridW ‘26
• Woohyeon Baek*, Jonghyun Bae*, Donghyun Lee, Hyunwoong Bae, Yeonhong Park, Jae W. Lee. “Liquid: Mix-and-Match Multiple Image Formats to Balance DNN Training Pipeline”. APSys ‘23
• Jonghyun Bae, Woohyeon Baek, Tae Jun Ham, Jae W. Lee. “L3: Accelerator-Friendly Lossless Image Format for High-Resolution, High-Throughput DNN Training”. ECCV ‘22