Alex Fizer

Designed and built a responsive personal website with a polished one-page layout, interactive visual elements, and easy-to-update content sections.

Developed a self-supervised computer vision pipeline to investigate how data augmentations influence the information captured by neural-network embeddings. I generated a dataset of 64,000 synthetic images with randomized shape, color, rotation, size, and position.

Then, I trained two six-layer convolutional encoders. One model learned color-sensitive representations, while the other used color augmentation to prioritize shape. I evaluated the resulting two-dimensional embedding spaces through cluster visualizations and nearest-neighbor image retrieval.

Designed and built an ESP32-based system that automatically dispensed measured quantities of pre-workout powder and water. The system used an HX711 load-cell interface to monitor weight in real time and stop the pump or powder-dispenser motor after reaching a target amount. I developed the embedded C++ logic for the scale, LCD, physical controls, pump, and dispensing motor, and soldered the supporting control and power circuits across two perfboards.

Designed a reusable Java component that models a simplified artificial neuron using weighted input signals, threshold-based activation, and normalized output calculations. Created the interface, abstract secondary implementation, and concrete representation from an open-ended software design prompt using object-oriented programming and design-by-contract principles.

Led a three-person team to develop, manufacture, and test a prototype robot to clear residential roadways. Designed a complex series of circuits consisting of multiple DC and servo motors to ensure a final product capable of pushing a payload capacity of over 5 pounds. Created a program to interface with circuitry in C++ using Arduino IDE.