Akara Supratak
Assistant Professor at Faculty of Information and Communication Technology (ICT), Mahidol University
Bangkok, Thailand
Summary
🤩
Rockstar🎓
Top SchoolAkara Supratak is an Assistant Professor at Mahidol University with 11 years of experience bridging academic research and practical ML engineering, after PhD and MSc study at Imperial College London. His work focuses on training deep neural networks on large-scale biosignal, image, and text datasets, with notable contributions to sleep-stage scoring (DeepSleepNet) and single-image super-resolution (SRGAN) where he improved data pipelines and deployment readiness. He combines teaching experience across undergraduate courses with hands-on MLOps and model lifecycle management, streamlining data acquisition and removing heavy dependencies to ease deployment. Based in Bangkok, he brings a researcher’s rigor to production problems, often optimizing training workflows and data validation steps that are easy to overlook in academic codebases.
11 years of coding experience
Doctor of Philosophy (Ph.D.), Computing Research, Doctor of Philosophy (Ph.D.), Computing Research at Imperial College LondonBachelor's Degree, Computer Science, First Class Honor, Bachelor's Degree, Computer Science, First Class Honor at Mahidol University
English
Stackoverflow
9reputation
161reached
0answers
1question
Github Skills (15)
mask-rcnn10
faster-rcnn10
machine-learning10
deeplearning-ai10
lstm10
deep-learning10
cgan10
tensorflow10
python10
dcgan10
super-resolution10
image-processing9
neural-network9
dockers7
docker7
Programming languages (4)
Github contributions (5)
akaraspt/deepsleepnet
May 2017 - Nov 2020
DeepSleepNet: a Model for Automatic Sleep Stage Scoring based on Raw Single-Channel EEG
Role in this project:
MLOps Engineer
MLOps EngineerContributions:1 release, 1 review, 23 commits in 3 years 6 months
Contributions summary:Akara primarily focused on setting up and configuring the environment for training and deploying the DeepSleepNet model. They addressed minor code typos related to training epochs. They integrated the project with eAE and Tensorlayer frameworks, suggesting a focus on model deployment and lifecycle management. Additionally, the user updated the download links for the dataset and removed MongoDB dependencies, contributing to streamlined data acquisition and deployment processes.
pytorchsleep-stagescoringdeep-learninglstm
tensorlayer/SRGAN
Jun 2017 - Jun 2017
Photo-Realistic Single Image Super-Resolution Using a Generative Adversarial Network
Role in this project:ML Engineer
ML EngineerContributions:8 commits in 4 days
Contributions summary:Akara primarily contributed to the project by modifying scripts related to image data handling and model training within the context of single image super-resolution. They added a script to download image data from a specified URL file, including functionality to check image sizes and filter out low-resolution images. Furthermore, they made changes to the training script, adjusting learning rates, and logging, along with merging code from other sources. This suggests a focus on improving data preparation, model training, and the overall workflow for the SRGAN model.
adversarialsingle-image-super-resolutionrealistictensorflowvgg16
Find and Hire Top DevelopersWe’ve analyzed the programming source code of over 60 million software developers on GitHub and scored them by 50,000 skills. Sign-up on Prog,AI to search for software developers.