Dongseok Im

AI/ML System and Silicon Architect

About Me

I am currently working at Google in San Diego, CA, USA, as an AI/ML Systems and Silicon Architect.

I received the B.S. degree in Electrical Engineering from Pohang University of Science and Technology (POSTECH), South Korea, in 2018, and the M.S. and Ph.D. degrees in Electrical Engineering from the Korea Advanced Institute of Science and Technology (KAIST), South Korea, in 2020 and 2023, respectively. I was a Postdoctoral Researcher at the KAIST Institute of Information & Electronics from 2023 to 2025.

My research interests are (1) energy-efficient AI/ML system-on-chip design, (2) digital circuit-, architecture-, and software-level AI/ML accelerator design, and (3) custom silicon chip and demonstration system board development.

News

(Updated: June 2025) New Paper Accepted at OJ-SSCS

- The new journal titled “An Overview of AI Hardware Architectures and Silicon for 3D Spatial Computing Systems” is accecpted at IEEE Open Journal of the Solid-State Circuits Society (Early Access)

Abstract: As artificial intelligence (AI) advances, 3D spatial computing has emerged as a key application in various fields. It interprets the 3D space surrounding users and provides them with useful information. This paper presents a survey of AI hardware architectures and silicon solutions for 3D spatial computing systems. The survey categorizes five domains: 3D data capturing, 3D data analysis, 3D hand motion analysis, simultaneous localization and mapping (SLAM), and 3D rendering. Each session analyzes design considerations for domain-specific accelerators. Finally, the paper discusses a next-generation 3D spatial computing platform that integrates various functions of 3D spatial computing systems using AI technologies.

(Updated: Mar 2025) New Career

- Starting a new position as an AI/ML system and silicon architect at Google, San Diego, CA, USA

Past News

- (Updated: Feb 2025) New US Patent Application

- (Updated: July 2024) New Paper Accepted at MICRO-57

- (Updated: Feb 2024) New Paper Accepted at HPCA 2024

First Authored Publications

OJSSCS 2025

IEEE Open Journal of the Solid-State Circuits Society

June 2025

(New) An Overview of AI Hardware Architectures and Silicon for 3D Spatial Computing Systems

Dongseok Im, Gwangtae Park, Junha Ryu, and Hoi-Jun Yoo

Paper Link

MICRO 2024

IEEE/ACM International Symposium on Microarchitecture

November 2024

CamPU: A Multi-Camera Processing Unit for Deep Learning-based 3D Spatial Computing Systems

Dongseok Im and Hoi-Jun Yoo

Paper Link

HPCA 2024

IEEE International Symposium on High-Performance Computer Architecture

March 2024

LUTein: Dense-Sparse Bit-slice Architecture with Radix-4 LUT-based Slice-Tensor Processing Units

Dongseok Im and Hoi-Jun Yoo

Paper Link

Micro Journal

IEEE Micro

May-June 2023

A Mobile 3-D Object Recognition Processor With Deep-Learning-Based Monocular Depth Estimation

Dongseok Im, Gwangtae Park, Zhiyong Li, Junha Ryu, Sanghoon Kang, Donghyeon Han, Jinsu Lee, Wonhoon Park, Hankyul Kwon, and Hoi-Jun Yoo

Paper Link

HPCA 2023

IEEE International Symposium on High-Performance Computer Architecture

March 2023

Sibia: Signed Bit-slice Architecture for Dense DNN Acceleration with Slice-level Sparsity Exploitation

Dongseok Im, Gwangtae Park, Zhiyong Li, Junha Ryu, and Hoi-Jun Yoo

Paper Link

JSSC Journal

IEEE Journal of Solid-State Circuits

January 2023

DSPU: An Efficient Deep Learning-Based Dense RGB-D Data Acquisition with Sensor Fusion and 3-D Perception SoC

Dongseok Im, Gwangtae Park, Junha Ryu, Zhiyong Li, Sanghoon Kang, Donghyeon Han, Jinsu Lee, Wonhoon Park, Hankyul Kwon, and Hoi-Jun Yoo

Paper Link

HOT CHIPS 2022

IEEE Hot Chips Symposium

August 2022

DSPU: A 281.6mW Real-Time Deep Learning-Based Dense RGB-D Data Acquisition with Sensor Fusion and 3D Perception System-on-Chip

Dongseok Im, Gwangtae Park, Zhiyong Li, Junha Ryu, Sanghoon Kang, Donghyeon Han, Jinsu Lee, Wonhoon Park, Hankyul Kwon, and Hoi-Jun Yoo

PPT Link

COOL CHIPS 2022

IEEE Symposium on Low-Power and High-Speed Chips and Systems

April 2022

A Low-power and Real-time 3D Object Recognition Processor with Dense RGB-D Data Acquisition in Mobile Platforms

Dongseok Im, Gwangtae Park, Junha Ryu, Zhiyong Li, Sanghoon Kang, Donghyeon Han, Jinsu Lee, Wonhoon Park, Hankyul Kwon, and Hoi-Jun Yoo

Paper Link

ISSCC 2022

IEEE International Conference on Solid-State Circuits

February 2022

DSPU: A 281.6mW Real-Time Depth Signal Processing Unit for Deep Learning-Based Dense RGB-D Data Acquisition with Depth Fusion and 3D Bounding Box Extraction in Mobile Platforms

Dongseok Im, Gwangtae Park, Zhiyong Li, Junha Ryu, Sanghoon Kang, Donghyeon Han, Jinsu Lee, and Hoi-Jun Yoo

Paper Link, Demo Video, News

JSSC Journal

IEEE Journal of Solid-State Circuits

February 2022

A Pipelined Point Cloud Based Neural Network Processor for 3-D Vision With Large-Scale Max Pooling Layer Prediction

Dongseok Im, Sanghoon Kang, Donghyeon Han, Sungpil Choi, and Hoi-Jun Yoo

Paper Link

TCAS-1 Journal

IEEE Transactions on Circuits and Systems I: Regular Papers

October 2020

DT-CNN: An energy-efficient dilated and transposed convolutional neural network processor for region of interest based image segmentation

Dongseok Im, Donghyeon Han, Sungpil Choi, Sanghoon Kang, and Hoi-Jun Yoo

Paper Link

S.VLSI 2020

IEEE Symposium on VLSI Technology and Circuits

June 2020

A 4.45 ms Low-latency 3D Point-cloud-based Neural Network Processor for Hand Pose Estimation in Immersive Wearable Devices

Dongseok Im, Sanghoon Kang, Donghyeon Han, Sungpil Choi, and Hoi-Jun Yoo

Paper Link

ISCAS 2019

IEEE International Symposium on Circuits and Systems

May 2019

DT-CNN: Dilated and transposed convolution neural network accelerator for real-time image segmentation on mobile devices

Dongseok Im, Donghyeon Han, Sungpil Choi, Sanghoon Kang, and Hoi-Jun Yoo

Paper Link

3D Video Projects

- 360° 3D VR Videos

This 360-degree 3D VR video projects a single 360-degree video into perspective views corresponding to human eyes with 3D depth effects through AI technology. The left video is captured through a dual-fisheye lens camera, and the right video is synthesized by applying depth to an original left scene. This project is run on a single RTX2080Ti card. The synthesis of a 360-degree 3D VR video takes 1 second per frame without any AI network finetunning. Left and right rendering displays take a total of 50 ms of latency as a view changes. Play VR videos on a Google Cardboard or AR/VR devices.

- 3D VR Videos

This project generates 3D VR videos from a single video through AI technology. The right video is synthesized from the left one fitting in different perspective views of human eyes. This project is run on a single RTX2080Ti card. The Synthesis of the right side video takes 0.3 seconds per frame without any AI network finetunning. Play VR videos on a Google Cardboard or AR/VR devices.

- 3D Glasses Videos

This project generates 3D Glasses videos from a single video through AI technology. Red and Cyan colors are separated from an original video, and pixels corresponding to Cyan colors are shifted based on a depth map for 3D effects. Play this video with Red-Cyan anaglyph glasses.

- 3D Photos

This project generates 3D photos from a single image through AI technology. This 3D effect provides a novel view synthesis that looks left and right.

Work Experience

- AI/ML System and Silicon Architect at Google

Mar. 2025 - Current

- Postdoctoral Researcher at KAIST Information & Electronics Research Institute

[Working Project] Designing high-performance hardware architectures for multi-camera AI-based 3D spatial computing systems

Sep. 2023 - Feb. 2025

Education

- Ph.D. in Electrical Engineering, Korea Advanced Institute of Science and Technology (KAIST)

Thesis Title: A Low-Power and Real-Time 3-D Object Recognition System-on-chip

Mar. 2020 - Aug. 2023

Advisor: Hoi-Jun Yoo

- M.S. in Electrical Engineering, Korea Advanced Institute of Science and Technology (KAIST)

Feb. 2018 - Feb. 2020

Advisor: Hoi-Jun Yoo

- B.S. in Electrical Engineering, Pohang University of Science and Technology (POSTECH)

Mar. 2014 - Feb. 2018

Awards & Scholarships

ISOCC Chip Design Contest Dest Poster Award, 2022

Microsoft Research Asia Fellowship Nomination Award, 2022

IEEE Solid-State Circuits Society ISSCC Student Travel Grant Award, 2022

28th Samsung HumanTech Paper Award (Circuit Design) Bronze Prize, 2022

Activities

[Invited Talk] Samsung AI Forum 2024 Poster Session Presenter, 2024

[Invited Talk] Apple Neural Engine Team Talk (San Jose, CA), “CamPU: A Multi-Camera Processing Unit for Deep Learning-based 3D Spatial Computing Systems,” 2024

[Invited Talk] Google TPU Architect Team Talk (San Jose, CA), “Circuit, Architecture, and Software Co-design of a Bit-slice Architecture for Dense and Sparse DNN Accelerations,” 2024

[Invited Talk] Samsung AI Forum 2023 Poster Session Presenter, 2023

[Invited Talk] Samsung Electronics DRAM Design Team Talk, 2023

[Invited Talk] Apple Neural Engine Team Talk (San Jose, CA), “Sibia: Signed Bit-slice Architecture for Dense DNN Acceleration with Slice-level Sparsity Exploitation,” 2023

[Invited Talk] Apple Neural Engine Team Talk (San Jose, CA), “DSPU: A 281.6mW Real-Time Depth Signal Processing Unit for Deep Learning-based Dense RGB-D Data Acquisition with Depth Fusion and 3D B-Box Extraction,” 2022

[Consultation] Expert Consultation on IITP ICT R&D Projects,2023

[Journal Reviewer] Transaction on Circuits and Systems II, 2025

[Journal Reviewer] IEEE Transactions on Very Large Scale Integration Systems, 2024

[Journal Reviewer] Elsevier Knowledge-Based Systems, 2024

[Journal Reviewer] Nature Scientific Reports, 2024

[Journal Reviewer] IEEE Journal of Solid-State Circuits, 2023

[Journal Reviewer] Springer The Journal of Supercomputing, 2023

[Journal Reviewer] IEEE Transaction on Circuits and Systems I, 2021, 2022

[System Demo] ACM/IEEE ISLPED Design Contest Presenter, 2022

[System Demo] IEEE ISSCC Demonstration Session Presenter, 2022

[International Exchange Program] Summer Session, UC Berkeley, 2016

Patent

1. “STEREOSCOPIC 360-DEGREE IMAGE/VIDEO GENERATION AND 3D DISPLAY SYSTEM,” United States Patent Application

2. “APPARATUS AND METHOD FOR GENERATING SIGNED BIT SLICE, SIGNED BIT SLICE CALCULATOR, AND ARTIFICIAL INTELLIGENCE NEURAL NETWORK ACCELERATOR TO WHICH THE SAME IS APPLIED,” United States Patent Application

3. “CONJUGATE GRADIENT ACCELERATION APPARATUS USING BAND MATRIX COMPRESSION IN DEPTH FUSION TECHNOLOGY,” United States Patent Application

4. “3D POINT CLOUD-BASED DEEP LEARNING NEURAL NETWORK ACCELERATION APPARATUS AND METHOD,” United States Patent Application

5. “스테레오스코픽 360도 이미지/영상 생성 및 3차원 디스플레이 시스템”, Korea Patent Application

6. “부호형 비트 슬라이스 생성기 및 그 방법과, 부호형 비트 슬라이스 연산기와, 이들을 적용한 인공지능 신경망 가속장치”, Korean Patent Application

7. “3D 포인트 클라우드 기반 딥러닝 신경망 가속 장치 및 그 방법”, Korean Patent Application

8. “뎁스 퓨전 기술에서의 띠행렬 압축을 이용한 켤레기울기 가속 장치”, Korean Granted Patent, 10-2600979