Sreejaya T

Research professional with 2 years of experience in machine learning and deep learning, specializing in multispectral image processing, model development and finetuning, and scalable ML pipelines for scientific datasets. Expanding expertise in Natural Language Processing and Large Language Models alongside computer vision.

Project Associate 1, Data Science and Machine Learning Team - Council of Scientific and Industrial Research - Fourth Paradigm Institute - Bengaluru

(2023-12)

Ministry of Science and Technology, Project: Region Specific Smart Agrotechnologies for Enhancing Soil and Plant Health

• Engineered an end-to-end, state-of-the-art pipeline to analyze high-resolution orthomosaicked UAV optical imagery for paddy disease detection, integrating acquisition and preprocessing for up to 11,000 images.
• Processed georeferenced raster datasets (GeoTIFFs) using different techniques aligning with various real world field disease scoring techniques for downstream model development.
• Ensured quality assurance of processed imagery through visual inspection, statistical checks and reproducibility documentation.
• Fine-tuned a Convolutional Neural Network (CNN) baseline with a set of hyperparameters for disease prediction of crop area with a validation accuracy of upto 75.31% and further fine-tuned pretrained CNNs, hybrid CNN-ViT architectures, Vision Transformer models on domain-specific agricultural datasets for disease detection to achieve test accuracies of upto 89%.
• Collected and curated agricultural field data using various handheld instruments such as SPAD, NDVI, Licor600 Porometer/Fluorometer etc over 4 fields across South India.

Integrated MSc - Physics with Minor in Scientific Computing - Amrita Vishwa Vidyapeetham, Deemed University (2018-07 - 2023-07)