Aditya
Swain

An advanced, secure, and intelligent online examination platform built with Flask and Python. This application is designed to conduct objective, subjective, and practical coding tests while ensuring academic integrity through state-of-the-art AI-based proctoring and monitoring.

An all-in-one web-based computer vision studio powered by Python, FastAPI, and OpenCV. Features 30+ advanced image processing filters, dual-pass AI inpainting/object erasure, custom frame generation, high-precision canvas resizing, and a browser-based multi-track video timeline editor.

Complete NLP pipeline: raw text cleaning → tokenisation → stop-word removal → TF-IDF feature extraction → word embedding generation → deep neural network classifier with dropout and early stopping. Evaluated on accuracy, precision, recall, and F1-score across all classes. Fully documented for reproducibility.

Developed custom object detection pipelines on custom-labelled datasets using transfer learning with pre-trained CNN architectures. Applied extensive data augmentation (rotation, scaling, contrast adjustment) and regularisation strategies to improve model generalisation and significantly reduce overfitting.

Built an end-to-end image classification and landmark detection system using CNN architectures in TensorFlow and Keras — from raw data ingestion and preprocessing through to training, evaluation, and inference. Applied data augmentation and regularisation for production-grade generalisation.

End-to-end real-time pipeline: live video → Haar Cascade face detection → CNN-based identification → automatic MySQL timestamp logging. Data augmentation (rotation, brightness, flipping) ensured reliable performance under varied lighting and angles. Adopted for live operations at CTTC MSME.