Selected work in machine learning and cybersecurity.
A computer vision-based waste segregation system that uses image processing and machine learning to separate four types of waste: plastic, paper, can, or general. This project demonstrates the application of AI in environmental sustainability.
A CrewAI-based collaboration of four specialized agents — a data analyst, strategy developer, trade advisor, and risk advisor — that work through a hierarchical process to analyze a stock and produce a trading recommendation. Agents use live web search and site-scraping tools to gather market data, and take a stock ticker, capital, risk tolerance, and trading preferences as input.
Classifies Windows malware into 25 families by treating each executable as a grayscale image. An ImageNet-trained ResNet50 is fine-tuned on the Malimg dataset with a frozen backbone and a new classification head, using an 80/20 train/test split. After 10 epochs it reaches 95.2% training accuracy and 88.1% on the held-out test set.
A collection of my own walkthroughs and writeups for various cybersecurity challenges on TryHackMe. Includes detailed solutions and insights from Capture The Flag (CTF) competitions. Also features blogs on my journey to earning certifications like eJPT.