About Me
Learn more about my background and experience.
Hi, I’m Kaung Myat Kyaw, a Machine Learning Engineer with a strong foundation in building scalable, data-driven systems. I’m pursuing an M.S. in Computer Science at the University of Colorado Boulder (GPA: 4.0), where I focus on machine learning, deep learning, and big data systems. I’ve applied my skills across impactful projects—from deploying a lung cancer detection model with 99% accuracy to building the Flow State App, a productivity tool for students in Myanmar IDP camps. I also volunteer as a software engineer, improving platforms and tools that serve real-world communities. My toolkit includes Python, PyTorch, TensorFlow, Hugging Face, AWS, and MLOps practices for production-ready ML solutions. I’m passionate about using AI to solve meaningful problems and enjoy working on teams that value innovation, performance, and social impact.
Skills & Technologies
A comprehensive overview of my technical skills and the technologies I work with.
Work Experience
My professional journey and key accomplishments.
Developed Website and an App called Flow State for student in Myanmar IDP campus.
Education
My academic journey and educational background.
University of Colorado Boulder
Specialized in Artificial Intelligence and Machine Learning
Certifications
Professional certifications and achievements that validate my expertise.
Learned end-to-end ML production workflows including deployment strategies, CI/CD pipelines, monitoring, data drift detection, and scalable infrastructure for reliable ML systems.
Featured Projects
A showcase of my recent work and personal projects that demonstrate my skills and creativity.
Flow State App
Flow State App
Designed and deployed a full-stack productivity app combining Pomodoro timer, meditation, and task-linked focus cycles, tailored for displaced students in Myanmar IDP Refugee camps. Used static generation and cloud deployment with CI/CD via GitHub and Vercel. Supported 100+ users; aimed to enhance focus, engagement, and mental health using behavioral reinforcement loops.
Retrieval-Augmented Generation for AWS Exam Q&A
Retrieval-Augmented Generation for AWS Exam Q&A
Built a RAG-based system for AWS certification exam questions that eliminated the need for costly fine-tuning on extensive documentation. By creating a dataset with LLMs from presentation slides, the system avoided retraining with documentation updates. This solution improved performance by 24%, achieving a higher accuracy than fine-tuned Gemma3 models.
Customer Segmentation with Unsupervised Learning
Customer Segmentation with Unsupervised Learning
Segmented customers in an online retail dataset using K-Means and Hierarchical Clustering; applied EDA, outlier detection, transformation, and standardization; tuned hyperparameters using Elbow and Silhouette methods to identify meaningful clusters for business insights. Engineered features and removed outliers to boost cluster interpretability by 50%
Lung Cancer Detection (CT Image-Based)
Lung Cancer Detection (CT Image-Based)
Trained multiple supervised models on lung CT scan data to assist physicians in early cancer detection. Achieved 99% accuracy with SVM; documented full model evaluation in published report. Compared models using ROC and precision-recall; aimed to reduce clinical turnaround time
BBC News Classifier with Supervised and Unsupervised Models
BBC News Classifier with Supervised and Unsupervised Models
Built a news classification system trained on the BBC dataset using both supervised (SVC: 98.5% accuracy) and unsupervised (NMF: 95.5% accuracy) methods. Preprocessed and vectorized news articles to categorize into business, politics, tech, and sports.
Personal Portfolio Website
Personal Portfolio Website
Built and deployed a fully static, SEO-optimized developer portfolio site with custom project showcase and continuous deployment.
Get In Touch
I'm always open to discussing new opportunities, interesting projects, or just having a chat about technology.