Vishnu S

Developed a fully offline Android Skincare application in Kotlin, migrating a Flask REST API to native mobile and eliminating cloud dependency. Implemented LLM observability and engineered a real-time on-device processing pipeline using CameraX and Kotlin coroutines.

Designed and deployed multi-agent microservices using LangGraph, CrewAI, and FastAPI with a React/TypeScript frontend. Fine-tuned PyTorch models for industrial machinery anomaly detection, achieving 96% precision and reducing diagnostic review time by 30%+.

Built fault-tolerant ETL pipelines for processing 6M+ bilingual electoral records using Python, SQL, and Selenium. Automated data extraction from scanned documents using custom-trained CNNs, Tesseract OCR, and Amazon Textract, reducing manual processing effort by 50%+.

Published IEEE paper on deep learning-based Alzheimer's detection from MRI data, achieving 99.06% accuracy. Co-authored IEEE paper on brain MRI multi-class segmentation using CNN and Hugging Face transformer pipelines.

Developed a geospatial risk assessment ensemble model (XGBoost, LightGBM) in Python, processing over 500,000 historical records to forecast disaster probabilities with 89% accuracy. Designed and deployed an interactive Node.js and Express dashboard, reducing manual reporting time by 40%.

Grade: 3.78

Pursuing a Master of Science in Computer Science with a GPA of 3.78, expected to graduate in May 2026.

Grade: 8.94/10

Completed a Bachelor of Engineering in Computer Science and Engineering with a GPA of 8.94/10.