Projects

The AutoMat project is a research effort focused on developing compressed neural representations of physically-based materials for real-time graphics applications. Its goal is to drastically reduce memory storage requirements for rendering while maintaining high visual fidelity, making it suitable for video games and interactive 3D graphics. Core Capabilities Neural Texture Compression: Learns compact material representations to replace dense texture maps. Auto-Encoder Architectures: Implements and compares VQ-VAE, convolutional, and MLP-based models for compression. ...

The Slipstream project is a Formula 1 database visualizer designed to make race data, driver statistics, team performance, and circuit information more accessible and engaging. Built on PostgreSQL with interactive visualizations, it provides fans and analysts with data-driven insights into Formula 1 history and performance. Core Capabilities Race & Season Exploration: Search and view details of past races by year and venue, including winners, podium finishers, and full results. Driver & Team Stats: Look up performance statistics such as wins, podiums, points, and season rankings. Interactive World Map: Explore global F1 circuits plotted on a map with coordinates, tooltips, zoom, and click-to-navigate functionality. Position Distribution: Visualize how often a driver finishes in each position across all races, highlighting performance consistency. Circuit Profiles: Dive into track-specific insights such as historical winners, top constructors, and all past races at that venue. Technologies Used PostgreSQL: Backend database storing races, results, drivers, constructors, and circuit data. Python / Flask: Backend service for handling routes and database queries. D3.js: Frontend visualization library powering position distribution histograms and interactive circuit maps. HTML, CSS, JavaScript: For frontend rendering and user interaction. Database Design & Queries Driver Pages: Aggregate queries compute podium finishes, wins, laps driven, and best performances for each driver. Circuit Pages: SQL joins and grouping identify dominant constructors and drivers at each track, with historical race listings. Relaxed Constraints: Adjustments were made to account for real-world edge cases, such as drivers not finishing (NULL position) or starting from the pit lane (Grid = 0). Related Links GitHub Repository

The RenderMark project is a web-based application designed to render parsed Markdown text on video templates, providing a convenient platform for users to create and download customized videos. Functionalities Markdown Editor: Allows users to enter text and images using a Markdown editor. Image Conversion: Converts images pasted from the clipboard into links for efficient rendering. Template Management: Displays a list of available video templates for user selection. Markdown Parsing: Parses Markdown files to extract text and images for video rendering. ...

Covi-Track is a web application designed for tracking COVID-19 hotspots. Leveraging GeoIP2, Google Maps Javascript API, and the Django web framework, this tool provides an interactive and informative platform for users to stay updated on COVID-19 cases. Features The website predominantly features an interactive map where each marker represents a user. The markers are categorized based on the user’s COVID-19 status, such as: Contracted COVID-19: Users who have been diagnosed with COVID-19. Vaccinated: Users who have received the COVID-19 vaccine. Recovered: Users who have successfully recovered from COVID-19. This categorization provides a visual representation of COVID-19 data, allowing users to quickly grasp the current status of individuals in different categories across geographical locations. ...

ASCII Art Generator is a command line tool that empowers users to generate text in ASCII Art Fonts effortlessly. This tool utilizes the Commander framework to provide a seamless command line interface for quick and efficient ASCII art creation. Installation Open the directory containing the project npm i -g . Usage art help Help with specific commands: art help <command> Related Links GitHub Repository

A Fire Alarm System designed with Arduino technology to enhance fire detection capabilities. This system utilizes Gas and Temperature sensors to monitor environmental conditions and promptly identify potential fire hazards. The recorded data is seamlessly transmitted and displayed on an Android App, leveraging Firebase for secure and reliable data storage. Features Real-time Monitoring: Continuously monitors gas levels and temperature for timely detection of potential fire incidents. Android App Integration: Utilizes a dedicated Android App to display real-time sensor data, ensuring accessibility and convenience for users. ...

This project involves the implementation of a handwriting recognition system using a Convolutional Neural Network (CNN). The system reads images using the opencv-python library and utilizes a model defined in the following code: Data Preprocessing The dataset is preprocessed by splitting it into training and testing sets. The images are reshaped to a size of 28x28 pixels, and the corresponding labels are one-hot encoded. Convolutional Neural Network Model The CNN model is defined with three convolutional layers followed by max-pooling layers. It concludes with fully connected layers for classification. ...

This project delves into Kendrick Lamar’s decade-long musical career, specifically analyzing the themes and topics prevalent in his music and lyrics. The primary methodology employed for this exploration is Non-Negative Matrix Factorization (NMF). The aim is to gain insights into the evolution of Kendrick Lamar’s artistic expression over the span of his albums. Analysis Topic Extraction NMF is applied to uncover latent topics present in Kendrick Lamar’s lyrics. These topics represent recurring themes and subjects in his music. ...

A comprehensive visualization dashboard showcasing satellite data from the Union of Concerned Scientists (UCS). This interactive platform is designed to provide an immersive exploration of the satellite dataset, offering valuable insights into various parameters. Technologies Used The visualization is powered by Plotly and Dash, enabling us to create an engaging and responsive user experience. These cutting-edge packages ensure that you can interact seamlessly with the satellite data, gaining a deeper understanding of the information presented. ...

VTOP Course Registration Scraper is a Python script designed to scrape course registration information from the university website and save it into a CSV file. This extracted data can then be used for further integration with timetable creation software. Usage pip install -r requirements.txt Enter the required values in the config dictionary and add course codes of the required courses in the courses list Leaving the list empty will scrape all courses visible to you (BEWARE, YOU CAN GET SOFTLOCKED OUT FOR 15 MINUTES IF YOU DO THIS) python scrape.py Output A courses.csv file will be created. Related Links GitHub Repository