Gukyeong Kwon

Senior Applied Scientist @ Amazon Artificial General Intelligence

About Me

Hi, my name is Gukyeong Kwon. I am a Senior Applied Scientist at Amazon Artificial General Intelligence (AGI). I completed M.S. and Ph.D. in School of Electrical and Computer Engineering at Georgia Tech in 2018 and 2021, respectively, under the supervision of Dr. Ghassan AlRegib. My research interests are machine learning, computer vision, and multimodal foundation models.

Please check my CV and Google scholar profile for more information.

News:

Experience

Amazon AGI

Senior Applied Scientist

October 2023 - Present

  • Contributed to Amazon Nova Multimodal Foundation models which achieve frontier intelligence in diverse vision and language tasks.
  • Led the launch of Amazon Nova Multimodal Embeddings model, a state-of-the-art multimodal embedding model which is the first embeddings models to support five modalities as input (text, documents, images, video and audio) for agentic RAG and semantic search applications.

AWS AI Labs

Applied Scientist

January 2021 - October 2023

  • Developed Amazon Titan Multimodal Embeddings model which can perform accurate and contextually relevant vision language search.
  • Conducted research on vision and language representation learning. Masked Vision and Language Modeling (MaskVLM) was proposed and accepted for publication at ICLR 2023.

AWS AI Labs

Applied Scientist Intern

May 2020 - August 2020

  • Developed regularization techniques for vision and language models and achieved improved performance in visual question answering, caption-based image retrieval, and referring expressions.

Panasonic Automotive

Deep Learning Research Intern

May 2018 - July 2018

  • Developed deep learning-based algorithms for drivers’ misbehavior detection in autonomous vehicles.
  • Focused on driver’s pose estimation and hand detection algorithms using Tensorflow and C++.

Georgia Tech

Graduate Research/Teaching Assistant

January 2016 - December 2020

  • Developed an anomaly detection algorithm using gradient-based representations which characterize knowledge that the model has not learned.
  • Conducted research on accident event detection algorithms to detect abnormal situations in driving scenarios and ensure safe autonomous driving.
  • Introduced a large-scale traffic sign recognition dataset for robust visual understanding under challenging conditions.
  • Developed a perceptual video quality assessment (VQA) metric which achieved the state-of-the-art performance in estimating the impact of visual distortions on human perception.

Publications

Amazon Artificial General Intelligence, “Amazon Nova Multimodal Embeddings: Technical Report and Model Card,” Tech report, 2025.

[Amazon Science]

Amazon Artificial General Intelligence, “The Amazon Nova Family of Models: Technical Report and Model Card,” Tech report, 2024.

[Amazon Science]

G. Kwon, Z. Cai, A. Ravichandran, E. Bas, R. Bhotika, and S. Soatto, “Masked Vision and Language Modeling for Multi-modal Representation Learning,” International Conference on Learning Representations (ICLR), 2023.

[arXiv]

Z. Cai, G. Kwon, A. Ravichandran, E. Bas, Z. Tu, R. Bhotika, and S. Soatto, “X-DETR: A Versatile Architecture for Instance-wise Vision-Language Tasks,” In Proceedings of the European Conference on Computer Vision (ECCV), 2022.

[arXiv] [GitHub]

G. Kwon and G. AIRegib, “A Gating Model for Bias Calibration in Generalized Zero-shot Learning,” In IEEE Transactions on Image Processing, 2022.

[arXiv] [GitHub]

G. Kwon, M. Prabhushankar, D. Temel, and G. AIRegib, “Backpropagated Gradient Representations for Anomaly Detection,” In Proceedings of the European Conference on Computer Vision (ECCV), 2020.

[arXiv] [GitHub] [Short Video] [Slides]

G. Kwon, M. Prabhushankar, D. Temel, and G. AIRegib, “Novelty Detection Through Model-based Characterization of Neural Networks,” 2020 IEEE International Conference on Image Processing (ICIP), Abu Dhabi, United Arab Emirates (UAE), 2020.

[arXiv] [GitHub] [Slides] [Video]

M. Prabhushankar, G. Kwon, D. Temel, and G. AIRegib, “Contrastive Explanations in Neural Networks,” 2020 IEEE International Conference on Image Processing (ICIP), Abu Dhabi, United Arab Emirates (UAE), 2020. (Top Viewed Special Session Paper Award)

[arXiv] [GitHub] [Slides] [Award]

G. Kwon*, M. Prabhushankar*, D. Temel, and G. AIRegib, “Distorted Representation Space Characterization Through Backpropagated Gradients,” 2019 IEEE International Conference on Image Processing (ICIP), Taipei, Taiwan, 2019, pp. 2651-2655. (* : equal contribution, Best Paper Award (top 0.1%))

[arXiv] [GitHub] [Poster]

M. Prabhushankar*, G. Kwon*, D. Temel, and G. AIRegib, “Semantically Interpretable and Controllable Filter Sets,” 2018 25th IEEE International Conference on Image Processing (ICIP), Athens, 2018, pp. 1053-1057. (* : equal contribution)

[arXiv] [GitHub] [Poster]

D. Temel, G. Kwon*, M. Prabhushankar*, and G. AlRegib, “CURE-TSR: Challenging Unreal and Real Environments for Traffic Sign Recognition,” MLITS workshop in Neural Information Processing Syste (NIPS), Long Beach, U.S.A, December 2017. (* : equal contribution)

[arXiv] [GitHub] [Poster]

M. A. Aabed, G. Kwon, G. AlRegib, “Power of tempospatially unified spectral density for perceptual video quality assessment,” 2017 IEEE International Conference on Multimedia and Expo (ICME), Hong Kong, 2017, pp. 1476-1481. (Finalist of the World’s FIRST 10K Best Paper Award (top 3%))

[arXiv] [GitHub] [Slides] [Award]