Side Projects

AI Model for Adaptive Traffic Control

Python, PyTorch, SUMO, MatPlotLib - Project Link

Built a Deep Reinforcement Learning agent to automate traffic lights across multiple scenarios.

Developed the agent using PyTorch Neural Layers to record inputs like possible direction, traffic congestion in current and next lane.

Used SUMO simulator with a Python connector to run train the agent and run simulations.

Utilised MatPlotLib to record metrics like average wait times, reward function, congestions to track success of the agent.

Cyclistic Data Analysis

Python, SQL, Excel, PowerPoint, Tableau - Project Link

This project was built as part of my Google Data Analytics Certificate Program.

Consolidated multiple CSV files into an SQL database for cleaning and ETL operations.

Cleaned data was exported to Python for statistical analysis, Excel for basic charts, and Tableau for complex geographical representation and analysis.

Prepared a final PowerPoint to aggregate generated insights and convey findings on customer use and suggest improvements to the business model.

S&P500 Dashboard

R, RShiny, DPLYR, RStudio, Plotly Project Link

Built a standalone platform to view stock prices and volume changes over a period. Can be expanded to use any stock and any data by changing the ticker.

Scrapes historical data from Yahoo Finance, can alternatively work with existing CSVs or alternate timelines and summaries.

Developed in RStudio and written completely in R with some add-on libraries to make scraping data easier.

Morse Code Game

ARM, C, RaspberryPi, CMake, Doxygen Project Link

Led a team of 5 to build a Morse Code with multiple levels and taking input from a Raspberry Pi.

Wrote the entire ARM code to retrieve '-' and '.' based Morse Code from the button on the Raspberry Pi.

Collaborated with a team member to implement string matching in C using a Trie inspired data structure.

Produced clear and useful documentation and functional flow charts using Doxygen and managed versions using Git

Blockchain based Emission Logger

Python, SQLite, Socket Programming Project Link

Developed the concept of an immutable emission logger based on blockchain technology that can be shared and validated via a P2P network.

Led a team of 3 individuals to develop this concept into project using Python for development, SQLite for database handling, and SHA256 for encryption.

Built a peer application to help new clients retrieve latest blocks and add new blocks to the decentralised blockchain.

Dynamic Blocks Sample Mount | AP Portfolio Lab