Featured Work
Building AI that makes an impact
End-to-end object detection and segmentation pipeline using Mask R-CNN to analyze seed germination patterns with 95% accuracy.
- Collected and annotated ~1,000 seed images with polygon masks
- Pipeline analyzes ~20 seeds per image simultaneously
- Achieved 95% accuracy and 98% F1-score
- Scalable architecture ready for deployment
Sophisticated rice classifier using metric learning to separate visually similar varieties in embedding space, achieving 97% accuracy.
- ~1,500 images per variety across multiple rice types
- Fast R-CNN detector for seed localization and cropping
- Metric learning pipeline for embedding space separation
- Binary classifier (EfficientNet + ResNet) at 97% accuracy
Other Projects
AI Mental Health Chatbot
Empathetic chatbot for students using NLP and sentiment analysis to provide accessible mental health support.
Viral Evolution Research
Analyzing SARS-CoV-2 coinfections and viral evolution patterns using NGS data and bioinformatics pipelines.
Blog
Coming Soon
Building Annotation Pipelines at Scale
Lessons from annotating 1000+ images for computer vision models.
Coming Soon
Metric Learning in Practice
How embedding spaces solve fine-grained classification problems.
Coming Soon
From Research to Production
Deploying computer vision models in real-world environments.
