Completed research work on malware image classification, in which I got leverage from transfer learning using various ImageNet models, and then extracted features by freezing the fc.layers of neural network, and fed our images data, then using Particle Swarm Optimization I optimized that feature vectors, and then finally classified and compared results of all models and classifiers (SVM & KNN), then using Bayesian Hyperparameters Optimization I optimized hyperparameters to see the difference and used Explainable AI to increase transparency and have meaningful impact.

Trained an AI Model using random forest tree, to detect anomalies in connection logs. The model was trained using Zeek Connection Logs metadata, which includes IP (Src. & Dest.), connection type, request protocols, no. of requests, RTT, checksum etc.

Torrent Control App allows users to remotely start, stop, and manage torrents through a web interface while their laptop handles the downloading. Built with Flask for the backend and a web interface using HTML, CSS, and JavaScript, it runs over a local network for seamless torrent control.