Yar Zar Moe Myint

Migraine Prediction Mobile Application — Machine Learning + Android(Kotlin) (Junction 2025 Hackathon)

• ML Pipeline:

  • Generated realistic synthetic physiological data (workload, stress, HRV) with custom behavioral + circadian mathematical models. Built and trained an LSTM time-series model to predict next-hour migraine probability. Implemented model explainability with Integrated Gradients to identify key contributing factors.

• Backend Infrastructure (Node + Express + PostgreSQL):

  • Created a scheduled prediction engine delivering three rolling 24-hour predictions. Designed database schema for user history, feature windows, and predictions. Integrated Gemini AI for recommendation generation based on risk factors.

• App (Android-Kotlin):

  • Built dashboard showing migraine probability, causal factors, and personalized advice. Enabled push notifications for rising migraine risk.

• Deployment:

  • ML model hosted on Render with scalable inference API. Backend deployed on Render with environment-based Firebase auth and secure secret handling.

JAN(2025)- PRESENT - Working part-time as Full-Stack Developer.

• Developed a cloud-based identity verification system that integrated mobile image capture, multi-factor verification (ID, face, address, email, and SMS), and blockchain-based digital signing of user data. 
• Leveraged AWS services (Lambda, RDS, Rekognition,S3) and Python/Django to create a scalable, secure backend for validating user-provided data. For Front-end, used React, Typescript/Javascript, HTML, CSS.

AUG(2021)- APRIL(2025) - Worked as software engineer (Test Automation Engineer, Android Developer, Embedded Engineer) in Elektrobit. (https://www.elektrobit.com/).

• Android Automotive & Middleware:

  • Developed native Android Automotive OS navigation applications using Jetpack Compose, Kotlin, MVVM, LiveData, and Room; followed test-driven development practices.

  • Integrated REST APIs with Retrofit and improved codebase with Coroutines and Flow.

  • Integrate to CI/CD pipeline using Github Actions.
  • Ported existing components to Android, Android-based OS, and embedded Linux (e.g., Raspberry Pi).
  • Built and modified components of the Android Open Source Project (AOSP) for Automotive OS.

• Test Automation & C++ Software Engineering:
  • Developed test-automation scripts using Python and Robot Framework.
  • Contributed to C++ and Android software development initiatives.
  • Prepared proofs-of-concept for new middleware applications

NOV 2024 –Independent Game Developer.Published 2D Pixel Art RPG Card Game on Steam Platform. (https://store.steampowered.com/app/3299520/King_of_Cards/)

• Developed a 2D pixel art RPG card game using Godot Engine with GDScript.
• Created and animated game assets using Aseprite.
• Managed full-cycle development from concept to publication on Steam.

JUN (2017) - DEC(2018) - Worked as a front-end developer in a company called CODIUM (https://www.codium.co/home). 

• Designed and implemented user interfaces (UI) using Angular, JavaScript, and TypeScript.
• Collaborated on user experience (UX) design and rapid prototyping.

JUN-AUG 2015   Intern at VGL lab,AIT, Thailand.

• Built the motorcycle helmet detection system that used OpenCV. The system detect whether the people  wore helmet or not from CCTV.
• It used OpenCV,SVM, HOG and Haar Cascades. The accuracy rate was 87%.

NOV 2015 - JAN 2016 Student Assistantship in AIT, Thailand. Helped developed a project called red light trafficking. 

Design of a User Interface in Telepresence Robot Navigation 

• User able to interact in VR for navigation while telepresence robot scan the real world environment and able to navigate with/without minimap.
• Used Unity Engine with C# for development in VR.

Progressive Web App Store (PWA)

• https://pwa-store.herokuapp.com/
  

Automatic Text Summarization 

• https://gitlab.com/YarZarM/text-summarization

ToDo App

• https://github.com/YarZarM/Todo

Gender Classification Using Gabor PCA and SVM.

• Developed a gender detection system from webcam and video file. The system uses Haar Cascade, Gabor features, PCA and SVM. The accuracy rate was 51%.