Portfolio

ChronicLog - Android Health Tracking Application

• Designed and developed android application with MVVM architecture for users with chronic illness that tracks medications, symptoms, and remediations with chat and provider share features

Reddit Client Android Application

• Worked with REST API integration via Retrofit and Gson
• Implemented real-time search filtering with MediatorLiveData, image loading with Glide, and fragment navigation with Android Navigation Component
• Features include subreddit browsing, post favoriting, and incremental search with highlighted results

HRIS Application

• Developed a full-stack Human Resources Information System streamlining onboarding and HR processes for 50+ employees
• Maintained the data model layer with ASP.NET Core, Entity Framework Core, and PostgreSQL, supporting scalable employee data management
• Designed and refined Employee entity models, creating EF Core migrations for schema updates
• Configured and debugged containerized environments using Docker Compose, troubleshooting PostgreSQL connectivity issues

SuperTux Learning Model

• Built and trained PyTorch models that learn to play TuxCart driving game achieving 96% classification accuracy
• Implemented convolutional neural networks (CNNs) for multi-task computer vision problems, including image classification, semantic segmentation, and depth estimation, using the SuperTuxKart dataset
• Designed a dual-head encoder–decoder architecture for simultaneous road boundary detection and depth prediction, using skip connections and multi-loss optimization (cross-entropy + L1 regression)
• Developed complete training pipelines in Google Colab, including data augmentation, GPU acceleration, and model checkpointing for reproducible experimentation

Hair Pulling Detector

Personal Project

• Built a computer vision tool using MediaPipe hand landmarks to detect hair-pulling behavior and alert the user in real time for increased self-awareness