A video surveillance system that allows authenticated and authorized users to access a system that recognizes faces from uploaded images and detect weapons (knives, guns, drones) in real time via connected camera feeds. It consists of the integration of a facial recognition model (InsightFace) and quantized object detection model (custom trained YOLOV5n model) into a webapp. This webapp also allows for real time SMS notifications, automatic logging for recognized faces and detected weapons with screenshots for future review, and a threat center that offers data visualisation from the logs of detected weapons.

A stock trading web application that caters to registered users by giving access to comprehensive information about publicly listed companies in the US stock market allowing for deposits, normal trade and basic options trading (long & short) of stocks with leverage seamlessly and securely through the integration of an authenticator for in-app transactions and third-party APIs, namely Tiingo for market data and Braintree for secure payment (deposit) processing. This platform was made to be ACID complaint with view endpoints secured with robust decorators and API endpoints secured with JWT. The transaction engine was implemented in C++ for high-performance and scalable processing.

Engineered a Python-based algorithmic trading system that executed a mean reversion strategy on commodity assets by integrating directly with the MetaTrader 5 platform API. Implemented real-time trade management and monitoring logic with terminal updates to track live performance and system status. Developed Python scripts to conduct in-depth time series analysis tests, including Bollinger bands analysis, Augmented Dickey Fuller (ADF), cointegrated ADF and Hurst Exponent tests on selected commodities. within Jupyter notebooks to validate strategy effectiveness and inform system parameters.

This project enables registered users and patients to schedule appointments with doctors, manage prescriptions through the pharmacy, and access tailored services like an AI-generated health report summary with lifestyle recommendations based on recorded patient vitals. The platform incorporates role-based access control, ensuring doctors, pharmacists, and front desk staff have secure access to role-specific functionalities. Additionally, the app leverages Celery and RabbitMQ to facilitate efficient asynchronous task processing, enhancing performance and scalability. API endpoints secured with JWT were also created to allow for CRUD operations.

Development and application of a suite of machine learning models from scratch to solve diverse problem types, which includes predictive modelling with the use of univariate and multivariate linear regression models for quantitative forecasting. Development of classification systems implemented via logistic regression and a multi-layer perceptron (MLP) classifier for binary and complex pattern recognition. Unsupervised learning is done by applying K-means clustering to uncover hidden patterns and groupings within unlabeled datasets.

A repository of C++ applications demonstrating core programming principles, including a console-based to-do list manager, a stock transaction engine, basic multithreading scripts, and utilities.