Curated Projects.

Built as a part of my Final Year Thesis. The aim of the project was to decrease the reliance on hardware for automated traffic management. The algorithm relies on available GPS data from services like Google Maps for lane and traffic density. THe information is then processed via a Neural Net, training the Multi-Agent model using reinforcement learning techniques to minimise metrics like average wait times, and average queue length. The model was built using Python and PyTorch training the model from base. The simulation was ran on SUMO across 7 maps of varying complexity to track the success of the algorithm. Achieved a First Class Honours on this work.

Technical Project completed as part of the Google Data Analytics Certificate. The project demanded collection of yearly data in CSV format. The data was (E)xtracted, (L)oaded into an MSSQL database, and (T)ransformed for consistency and correctness. The data was analysed via Python scripts and Tableau visualisations to analyse daily, weekly, and monthly trends in data drawing insights and forming recommendations. The summarised data was loaded and visualised in Excel and exported to PowerPoint for the final presentation.

Group project completed as part of the Computer Networking Module at Trinity College Dublin. Leveraged a Raspberry Pi on a LAN to connect multiple devices to a blockchain. Demonstrated usecase of tracking car emissions on a blockchain to for public access and strictness. Used Python multi-threading to handle concurrent collections, block hash secured with SHA-256, SQLite to store and retrieve standard emission rates for cars, and auto-update blocks after every trip.

Built a web scraper to scrape data in multiple Indian languages from Social Platforms like Facebook, Twitter, Google News, etc. Data was cleaned and stored in an MSSQL database to prepare to be fed into a LLM for processing and report the sentiment analysis of the public on specified topics. Used Python and Selenium to build the scraper handling data with standard REST APIs and browser scraping. Worked solo on the project developing a small GUI as well to aid the non-tech users in using the underlying technology.

Completed as part of a Freelancing job. Built a Python-based script (using only NumPy and Pandas) packaged with PyInstaller as a `.exe` to install on Windows and run independently. Consolidated data from 34 (files) x 5 (sheets) into 5 files. Reduced processing times from several days to minutes. Added features for after-processing to reduce even more human efforts in consolidating and cleaning data. Set up Excel macros to transform the final data into the required format, eliminating the need for manual work.

Pre-cursor to a template based website builder. Offers a look into standalone components and blocks built with Shadcn primitives and styled with Tailwind. The components are built for reuse and utilise props and dynamically genreates them to let the user extend the functionality without any code changes. Currently hosted on vercel, and offering previews only at the moment.