About
I’m Sai — a computer vision and remote sensing engineer based in Thailand.
I completed my Master of Engineering in Information and Communications Technologies at the Asian Institute of Technology (RSGIS, GPA 3.0/4.0) and hold a Bachelor of Engineering in Computer Science from Jawaharlal Nehru Technological University, Hyderabad (GPA 3.2/4.0).
Research
My thesis focused on detection and canopy size estimation of oil palm trees using deep learning–based instance segmentation (YOLOv8, YOLOv11, SAM) on multi-GSD UAV imagery — building end-to-end pipelines for precision agriculture and sustainability monitoring.
Other research interests include:
- Angular-temporal interaction for advanced pixel-level visual tasks
- End-to-end real-time object detection on static and dynamic imagery
- Federated learning for generative AI
- Adversarial defense against AI-generated steganography
- Multi-modal alignment gaps in generative AI
What I Work With
- Computer Vision — YOLO family, SAM, Mask R-CNN, instance segmentation, object detection
- Remote Sensing — UAV imagery processing, multi-GSD analysis, canopy metrics, geospatial ML
- ML/DL Stack — Python, PyTorch, Ultralytics, OpenCV, GDAL, QGIS, ArcGIS, SNAP, Roboflow
- AI Research — Text-to-image, text-to-video, prompt engineering, LLM agents
- Web & Tools — Jekyll, GitHub Pages, Obsidian, WordPress, Canva
- Currently learning — Neuroscience, Deep Reinforcement Learning, Applied Mathematics
Beyond the Screen
I’m an IPF-registered competitive powerlifter under TAAP/IPF. In 2025, I competed in both NQ1 (March) and NQ3 (November) in the U74kg Men’s Raw Open category — totaling 390 kg and 430 kg respectively, placing 3rd in both events. I also serve as Head of Events for powerlifting at AIT.
I believe training the body is as important as training the mind.
What This Blog Is About
This is where I write about things I’m learning, building, and thinking about — computer vision, AI tools, web development, and the intersection of research and real-world engineering. No frameworks, no fluff. Just pure HTML and CSS.