I'm Rathisan Rajakrishnan

Full-stack, location-aware web application designed to help University of Waterloo students identify available classrooms for studying in real time. The platform integrates official university scheduling data with live geolocation to determine room availability and dynamically rank study spaces based on proximity and current time. A custom Flask-based backend ingests, normalizes, and filters classroom data into actionable availability states, while handling time-sensitive logic to reflect real-world usage patterns. The frontend, built with Next.js and Mapbox, provides an interactive campus map and synchronized list view, enabling users to seamlessly explore open spaces and drill down into room-level schedules. By combining spatial data, temporal reasoning, and responsive UI design, Openr transforms fragmented scheduling information into a centralized, intuitive tool for campus navigation and study planning.

Lightweight Chrome extension designed to make mathematical content on the web more accessible by allowing users to instantly extract and copy LaTeX from KaTeX and MathJax-rendered equations. The extension uses content scripts to detect equation elements, parse underlying annotation data, and retrieve LaTeX expressions through DOM inspection, enabling one-click copying via the Clipboard API. A custom hover-based overlay interface was built with CSS to feel intuitive and unobtrusive, ensuring seamless interaction without disrupting page layout or performance. HoverTeX emphasizes usability, cross-site compatibility, and clean client-side architecture, making it easy for students and researchers to reuse mathematical notation directly from the browser.

Built a machine learning pipeline to assess mesothelioma risk from clinical data using an artificial neural network implemented in Keras. The work involved preprocessing and normalizing patient records with scikit-learn to ensure consistent feature representation and stable model training. Emphasis was placed on evaluating model performance and improving reliability through thoughtful data preparation and iterative tuning, resulting in a robust binary classification system for medical risk prediction.

A secure password manager web application that enables users to store and retrieve website login credentials through a personalized PIN-based access system. Users set a PIN during initial setup, which is required to unlock and view saved credentials, adding an extra layer of protection beyond standard authentication. The application supports securely storing usernames and passwords for multiple websites and retrieving them on demand, with encryption used to ensure sensitive data remains private. The focus of the project is on building a simple, user-friendly interface while applying core security principles to protect user information.

Developed a web-based encryption tool that allows users to encrypt and decrypt text using classic substitution ciphers, including the Caesar Cipher and ROT13. Users can customize shift values for encryption, while the application automatically determines the appropriate transformations for decryption. Each encrypted message is associated with a generated decryption key, enabling secure retrieval of the original text at a later time. The project emphasizes implementing core cryptographic concepts in Python while exposing them through a simple, interactive web interface.