Debanjan Chowdhury

Optimized real-time CenterNet object detection for Sony Alpha cameras, reducing false positives to <0.5%. Developed audio-visual lip-sync, multilingual speech synthesis, and AI-powered teaser generation pipelines.

Designed courses on Python, SQL, ML, DL, and Computer Vision. Conducted classes and provided project guidance with live doubt clearing sessions.

Developed deep learning models for crop classification using PolSAR data and near real-time weed detection using YOLOv12, contributing to precision agriculture.

Developed a methodology for detecting Giant Rat Tail grass using UAV imaging and machine learning, achieving 95.5% accuracy with XGBoost.

Built an end-to-end AI review aggregator scraping 2,000+ reviews daily, implementing GPT-3.5 classification, sentiment analysis, and fake review detection with MySQL storage and dashboards.

Developed Random Forest tooth decay detection, YOLOv7 screw hole detection, plant leaf disease detection, and AI-powered intro/outro detection systems, improving efficiency and reducing costs.

Developed AI-powered database attribute categorization and video-to-PDF conversion tools, achieving high accuracy and significant reductions in manual effort and processing time.

Pursued a Master of Science in Data Science, focusing on advanced data analytics and machine learning techniques.

Earned a Bachelor of Technology in Computer Science and Engineering, building a strong foundation in software development and computer systems.