I am an associate professor at the graduate school of data science of Seoul National University (SNU). My main research area includes computer architectures, compiler and runtime technologies, design space exploration, and parallel programming, focusing on designing extensible and efficient software stacks for modern heterogeneous systems ("Systems for ML"), and applying data-driven approaches to solve optimization problems of computer systems ("ML for Systems"). 

I received my bachelor’s degrees in literature and computer science from Seoul National University, M.S. in computer science from UC San Diego (advised by prof. Michael Taylor), and Ph.D. in computer science from the University of Illinois at Urbana-Champaign (advised by prof. Sariva Adve). I worked as a research staff member at IBM T. J. Watson Research Center from 2015 to 2019. Before joining SNU, I was an assistant professor at the department of computer science and engineering at POSTECH (Pohang University of Science and Technology). 

I am leading Data-driven System Optimization (DSO) lab (tentative) at SNU. 

I am actively looking for outstanding graduate and undergraduate students and postdoctoral researchers for the projects listed below. When you contact me, please include an up-to-date CV and what project(s) in our group you are interested in. 

My CV can be found here. 

Current Research Projects

Machine learning for compiler optimizations: Generating high-performing codes for increasingly heterogeneous hardware calls for more flexible and adaptive ways to model program behaviors with different optimization decisions than traditional heuristic-based cost models. We take data-driven approaches where the code-performance relationship is accurately learned from code representation and profiling results.  CogR (PACT ’19) guides the OpenMP runtime scheduler by predicting whether an OpenMP target region will execute faster on CPU on GPU using a deep-learning based predictor model, while MetaTune extends an auto-tuning framework in a deep-learning compiler, TVM, to reduce autotuning overheads and generate better-optimized codes for tensor operations. Most recently, One-Shot Tuner (CC ’22) showed how online auto-autotuning overheads can be practically eliminated with a NAS-inspired performance predictor model trained with a small set of samples (open-sourced). 

Compiler technologies for emerging architectures: With the end of Dennard scaling, we are witnessing a major shift in the computer system and microarchitecture design towards exploiting more specialized and lightweight “accelerators” of different types instead of relying on mostly general-purpose processors. Such heterogeneous systems pose an unprecedented challenge for the entire software stack to provide programmability and portability while delivering performance. We work on rethinking compiler and runtime technologies for heterogeneous systems with emerging architectures such as neural processing units (NPU) and compute-augmented memory (NDP/PIM).

PIMFlow (CAL ’22, CGO ’23) proposes software layers specifically designed to accelerate compute-intensive convolutional layers on PIM-DRAM. Integrated with the TVM compiler, PIMFlow provides graph-level optimizations to create more inter-node parallelism so that layers can execute parallel on GPU and PIM. While PIMFlow is about inference time performance, XLA-NDP (CAL ’23) focuses on enabling GPU-PIM parallel execution during model training (open-sourced). 

OpenCL compiler and runtime support for next-gen supercomputers: Modern supercomputers harness massive parallelism provided by host CPU’s and specialized computing elements such as GPU’s and NPU’s. In collaboration with ETRI and KISTI, we work on building Korea’s own next-generation supercomputers and providing OpenCL programming support with optimizing compilers and runtime.  

Accelerator design space exploration: DSE is a key research topic for application-specific accelerators, but has not been extensively explored in the context of processing-in-memory hardware design.  Our recent paper (ICCAD '23) introduces a heterogeneous analog computing-in-memory architecture that supports multiple tile and subarray sizes (“big-tile, little-tile”) with an end-to-end design space exploration tool to optimize latency, power, and area at the same time (open-sourced).