Summary
Overview
Work History
Education
Publications
Websites
STRENGTHS
Timeline
Generic
SOE SANDI HTUN

SOE SANDI HTUN

AI Engineer (Deeplearning Researcher)
Seoul,Dongjakgu

Summary

Deep learning researcher specializing in computer vision with a proven track record of developing cutting-edge models for object detection, image segmentation, and anomaly detection in CCTV driving videos. Proficient in PyTorch, TensorFlow, and Python, with expertise in creating DeepStream plugins for seamless integration of deep learning algorithms. Trusted by management to lead research initiatives and ensure team effectiveness and efficiency.

Overview

7
7
years of professional experience
7
7
years of post-secondary education
2
2
Languages

Work History

AI Engineer (Deeplearning Researcher)

Pintel Co., Ltd.
03.2023 - Current
  • 1. Lead the development of cutting-edge real-time object detection and image segmentation systems to improve the company's visual data understanding and analysis capabilities.
  • 2. Leveraged generative AI technology to develop an image captioning model that generates descriptive captions in Korean to advance the company's AI-based content understanding and localization efforts.
  • 3. Developed automotive accident detection and prediction utilizing deep learning methodologies to create predictive models to enhance safety measures and mitigate road hazards.
  • 4. Played a pivotal role in the research and development of advanced algorithms for multimodal research, contributing significantly to the company's mission to promote intelligent event detection across diverse data streams and leverage AI for comprehensive insights and decision-making.

Research Student (Masters' Thesis)

Seoul National University of Science and Technology
03.2021 - 02.2023
  • 1. Researched in the field of motion detection and recognition, focusing on problems similar to those faced in Activity Net Challenges, and achieved state-of-the-art performance by leveraging advanced deep learning techniques.
  • 2. By exploring innovative methodologies for traffic accident detection, I developed of predictive models aimed at enhancing road safety and accident prevention.
  • 3. Demonstrated proficiency in literature review, hypothesis formulation, data collection, and statistical analysis by demonstrating a comprehensive understanding of research methodology and best practices.
  • 4. Presenting research results at academic conferences and seminars to effectively communicate complex technical concepts to diverse audiences and receive recognition for academic contributions.

Senior Mobile Developer

Tamron Technology
02.2020 - 02.2021
  • 1. Architected and implemented complex mobile solutions demonstrating expertise in iOS SDK, UIKit, Core Data and other essential frameworks to meet project requirements and ensure scalability and maintainability.
  • 2. Work closely with cross-functional teams including product managers, designers, and QA engineers to gather requirements, iterate on design concepts, and deliver feature-rich iOS applications that exceed user expectations.

Junior Software Developer

Get All Myanmar Co., Ltd.
12.2018 - 11.2019
  • 1. Full-stack web development for microfinance system + payment management
  • 2. Javascript, React js, node js, graphQL, and mongoDB were applied for development.

Education

Master's Degree - Computer Engineering

Seoul National University of Science And Technology
Seoul
03.2021 - 02.2023

Bachelor's Degree - Computer Science and Engineering

University of Information Technology
Yangon
12.2013 - 12.2018

Publications

  • TempoLearn Network: Leveraging Spatio-temporal Learning for Traffic Accident Detection 12/2023

            1. Developed a new model , the TempoLearn network, which outperformed existing state-of-the-art models by achieving a 16.5% higher incident localization score (AUC) on the largest and most complex Detection of Traffic Anomaly (DoTA) dataset. Traffic accident datasets are available.

             2. We demonstrate the effectiveness of the TempoLearn network through experiments on the Car Crash Dataset (CCD) and further validate its robustness and performance on different datasets.

             3. Demonstrate proficiency in advanced machine learning methodologies by leveraging multimodal learning techniques to improve model accuracy and generalization capabilities.

https://ieeexplore.ieee.org/document/10360840



  • AI-Based Modeling Architecture to Detect Traffic Anomalies from Dashcam Videos 10/2022

     2022 13th International Conference on Information and Communication Technology Convergence (ICTC)

            1. Demonstrated a state-of-the-art multimodal learning approach, integrating the convolutional learning method with a self-attention model to extract relevant features and model temporal occurrences.

            2. Demonstrated the effectiveness of the proposed architecture in accurately classifying segmented traffic accidents into predefined groups and highlighted its practical applicability in improving road safety and accident prevention.

            3. Provided valuable insights and recommendations for future research and implementation, emphasizing the commitment to continuous improvement and advancement in the field of AI-based traffic analysis and anomaly detection.

https://ieeexplore.ieee.org/document/9952473



  • Describing Environmental Information in Videos Using Machine Learning 10/2021

     2021 21st International Conference on Control, Automation and Systems (ICCAS)

            1. Developed a new approach to video captioning that incorporates important contextual information such as human actions, objects, location, time, and weather.

            2. Created a new dataset that supplements the existing MSVD dataset with environmental context labels, addressing a significant gap in prior video captioning research.

            3. Implemented a state-of-the-art machine learning model that combines R(2+1)D 3D CNN for video feature extraction with S2VT RNN for environmental information encoding and decoding.

            4. Adopted a sequence-to-sequence framework tailored for video analysis to ensure the model generates coherent and contextually appropriate captions.

            5. Validated that our model delivers competitive performance compared to existing video captioning models evaluating with BLEU, METEOR, ROUGE-L, and CIDEr.

https://ieeexplore.ieee.org/document/9649840

STRENGTHS

  • Proficiency
  • Adept at developing and implementing complex deep learning models and algorithms, and at processing and processing large-scale data sets to train and optimize AI models for a variety of applications, ensuring strong performance and accuracy.
  • Communication
  • Demonstrated ability to clearly and persuasively communicate complex technical concepts and research findings to diverse audiences, including peers, stakeholders, and industry experts, fostering collaboration and knowledge sharing in the fields of artificial intelligence and deep learning.
  • Multi-tasking
  • Trusted by manager to explore and explain new research avenues, while consistently meeting primary responsibilities with efficiency and excellence.

Timeline

AI Engineer (Deeplearning Researcher)

Pintel Co., Ltd.
03.2023 - Current

Research Student (Masters' Thesis)

Seoul National University of Science and Technology
03.2021 - 02.2023

Master's Degree - Computer Engineering

Seoul National University of Science And Technology
03.2021 - 02.2023

Senior Mobile Developer

Tamron Technology
02.2020 - 02.2021

Junior Software Developer

Get All Myanmar Co., Ltd.
12.2018 - 11.2019

Bachelor's Degree - Computer Science and Engineering

University of Information Technology
12.2013 - 12.2018
SOE SANDI HTUNAI Engineer (Deeplearning Researcher)