Duck Harvest

A cozy drag-and-drop game where ducks are the crop, and pesky vegetables keep getting in the way of the player’s harvest. Players may unlock appearances for their ducks and customize them, giving their quacking crop a unique aesthetic.

Highlights:

• Custom Shaders
• Optimized for Mobile
• Integrated Audio Systems
• Post-Processing Effects
• Weighted Spawn System using Batching

That Beaver Game

Networked multiplayer destruction sim that focuses on a fast-paced, parkouring beavers. Players can sprint, leap, wall-run, and slam their way through the city while wreaking havoc upon everything in their path.

Contributions:

• Advanced Character Controller
• Advanced Camera Effects
• Integrated Mesh Fracturing System
• Networked Multiplayer

Plasma Fusion

Multiplayer arena battler in which two Plasmos (players) must choose abilities, and use them to defeat their opponent. After each round, the loser gains a new ability, while the winner gains nothing but pride. The first to win 5 times claims victory.

Contributions:

• Dynamic Multiplayer Input System
• Ability System Architecture
• Elemental Stats / Interactions Architecture
• Custom Scalable Object Pooling

PolyPuck Networked Bots

Added infrastructure for networked connections using UDP sockets and protobuf. This allows people, as well as custom made bots, to play the game remotely.

Created and trained a machine learning model to play the game and improve against the game’s pre-existing AI.

This project is a proof of concept to creating an API to allow anyone to create their own bots, and pit them against one another in PolyPuck.

Chocolate & Machine Learning

An analysis of a dataset comprised of various chocolate bar ratings, ingredients used, manufacturing location, and more.

Process involved analysis of previous uses of the dataset, identifying an un-explored aspect, filtered the data during preprocessing, then used it for training.

The data was used to train five different machine learning algorithms to attempt to find which aspects of chocolate effect the rating most.

AI Pathfinding

As part of Cacophony, enemies were developed to chase down the player. These enemies’ pathfinding went through many iterations over the course of development.

Starting with A* pathfinding as a base, the enemies were able to path to the player, but struggled with knowing where other enemies were, let alone where they were going.

Many path weight and cost adjustments, such as adding a ‘Claimed Cost’ to pathed tiles, allowed for fluid enemies that use teamwork to trap the player.