Arunabh’s Projects

Predicting Wine Quality from Physical and Chemical Properties

Utilized various machine learning algorithms to predict and visualize wine quality for red and white wine. Methods used included: best subset selection, lasso regression, principal components analysis, LOESS, KNN, and decision trees. View our final report and code here.

Breaking Down the Grandmaster’s Strategy

Completed a data science/machine learning-based research project with 2 other students. Analyzed a dataset of 20,000 chess games to identify relationships between W/L ratio, match length, matchmaking, opening strategy, defeat strategy, and player ratings. Implementing logistic regression, linear regression, and time-series forecasting to make accurate predictions and visualizations on chess strategy using Python, SQL, and Tableau. View our final report here.


Developed a React web application with 3 other students. The application allowed users to publish and read smells in their local area to avoid locations during outdoor activities or enjoy certain outings to their liking. For functionality and utility, the application warns people about bad smells in an area (i.e. skunk roadkill, areas of a city to avoid on a walk/run) or recommend friends to visit an area with nice smells (i.e. a clean-smelling public bathroom/cherry blossoms on campus). This ReactJS project uses TypeScript, JavaScript, and Firebase. The sketch is displayed above. Click here to view the GitHub repository.


Collaborated with a team of 4 entrepreneurs, as Product Lead, to conduct customer discovery through customer interviews, formulate a business thesis, identify customer segments, and develop a smart anti-rotation bike lock product for the e-bike industry. Our initial design is displayed above. View our Final Pitch here and our Business Model Canvas Deliverable here.

Sudoku Solver

In this application, I developed a recursive Sudoku solving algorithm in Python to solve faster than traditional brute force by using backtracking. To generate randomized puzzles, I utilized a REST API to randomly generate and solve puzzles from difficulties in easy, medium, and hard. Click here to view the GitHub repository.

Cyclic Arbitrage

In collaboration with Cornell Blockchain, I helped develop a cyclic arbitrage bot to trade on Uniswap and Sushiswap by conducting research on the use of a Bellman-Ford algorithm to find negative cycles in arbitrage opportunities. The project can be found on the Cornell Blockchain GitHub.

Pathfinding Simulator

In this application, I used PyGame to implement the unweighted A* Pathfinding algorithm that finds the shortest distance between two given nodes 60x faster than Dijkstra’s algorithm. Click here to view the GitHub repository.

Sorting Algorithm Visualizer

In this program, I used Java Swing to implement three sorting algorithms, running through various time complexities. The visualization of merge, selection, and insertion sort to sort up to 200 unsorted data points. Click here to view the GitHub repository.

Online Debate Education

I worked with numerous educators to facilitate online services that provide access to high-level debate rounds on foreign and domestic policy. Topics covered domestic surveillance, global space operations, judicial processes, and international monetary policy. Check out our work here.