cv

MS in Artificial Intelligence

Gwangju Institute of Science and Technology

• Major in Artificial Intelligence

Summer International Visitor UG

University of California Berkeley

• Relevent courses
  • CS188 - Introduction to Artificial Intelligence
  • STAT 20 - Introduction to Probability and Statistics

BS

Gwangju Institute of Science and Technology

• Major in Electrical Engineering and Computer science
• Relevent courses
  • Data Structure
  • Introduction to Algorithms
  • Signals and Systems
  • Discrete Mathematics
  • Computer Vision

Research Internship

IRRLab @ Gwangju institute of Science and Technology

• Study a Diffusion Generative Model
  • Implemented the VAE, VQVAE, and DDPM to understand how diffusion generative models works. All the code I have implemented is opened via my Github.
• Inception-ized Diffusion Model
  • With own novel suggestion, I implemented the Inception module-like diffusion model. In the paper, I suggest a novel way to train the diffusion model. That is, building a structure in parallel diffusion with 2 different resolution. The codes are opened via my Github.

AI Software Engineer as Internship

Brainsoft Inc.

• ESPNet
  • Train speech recognition model with AIhub KSponSpeech dataset(1000hr). By researching the ESPnet and Speech Recognition(SR), learned that SR integrates the idea of ‘alignment’. SR networks can vary by how they define the method of alignment.
• Android App using Speech Recognition
  • Implemented Android speech recognition app using Google speech-to-text API (Google Cloud) with Java and Kotlin. Learned the requirement of a small model that can run on the device offline. Small on-device models are more efficient because of the cost of communication, computing delay, and computing power.
• RNNoise(Speech Enhancement)
  • Train the RNNoise model and integrate it to use it on hearing aids. My primary job was to decrease the delay of the model from 20ms to 10ms and change the sample rate from 48k to 16k with code analysis and research about psychoacoustics. Also, integrate the RNNoise model from Keras into PyTorch to increase the train speed.

al-folio

• A beautiful, simple, clean, and responsive Jekyll theme for academics.

• Nobel Prize in Physics
• Matteucci Medal

• Max Planck Medal