Minyoung Shin
Software Engineer @ Samsung Research
đź“Ť Seoul, South Korea
About
I am a software engineer specializing in computer architectures for efficient AI acceleration. At Samsung Research, I lead the development of a cycle-accurate NPU simulator for Samsung AI TV SoCs and a design space exploration framework for next-generation NPUs targeting large language models. Previously, I validated an Accelerator-in-Memory prototype at SK hynix and helped achieve successful product qualification.
Research Interests
I am broadly interested in data-centric computing paradigms—including in/near-memory computing, memory-centric architectures, and computational storage—that reduce data movement across the memory hierarchy. In addition, I focus on agile HW/SW co-design methodologies with a simulator-driven approach, using architectural simulators to accelerate iteration, enable rigorous pre-silicon validation, reduce time-to-market, and make the deployment of new architectures more flexible. Ultimately, my research goal is to redefine computer architectures for AI by shifting the paradigm from processor-centric to data-centric computing.
Experience
Software Engineer @ Samsung Research (2022 - Present)
Application Engineer @ SK hynix (2021 - 2022)
Education
B.S. in Electrical and Electronic Engineering @ Yonsei University (2016 - 2021)
Selected Projects
- Cycle-Accurate NPU Simulator (2023 - Present)
- NPU Design Space Exploration (2025)
- QEMU-Based Full-System Emulation (2024)
- DSP Kernel Optimization (2022 - 2023)
- GDDR6-AiM System-Level Verification (2022)
- B.S. Thesis: Modeling Tensor Core Microarchitecture for NVIDIA Turing with Experimental Features (2020)
Publications & Patents
- Patent Pending: "Efficient Synchronization Protocol and its Hardware Module in Parallel Computing Systems" (2025)
- "System Architecture and Software Stack for GDDR6-AiM." IEEE Hot Chips 34 (2022)
Additional Experience
- ROK Army & United Nations Peacekeeping (2018 - 2019)