AI software is becoming increasingly complex, and the computation capacity of hardware is unable to keep up. However, flexible hardware platforms, called Field Programmable Gate Arrays (FPGAs), can be leveraged to create solutions that accelerate AI software. TensorBake is a software compiler that aims to reduce the challenges surrounding the adoption of FPGA technology and provide developers with a way to convert their neural networks to code that can be deployed to, and accelerated by, an FPGA.

Every year thousands of search and rescue expeditions are performed, using many man-hours and risking volunteers' lives. SwarmSearch is a drone system that utilizes a network of autopilot drones, along with machine learning to detect and locate missing persons while being completely autonomous. We plan on developing a mesh network capable of supporting thousands of micro-drones equipped with thermal and visual imaging systems trained using machine learning to detect human heat signatures. SwarmSearch will be able to seamlessly scale allowing for autonomously searching large areas.

HUBibi is a DIY smart hub that allows a user to connect virtually any home IoT device that uses the ZWave, Zigbee, or Thread protocol. In addition to being cheaper than commercial alternatives, it also provides greater ease-of-use than other open-source solutions. Rather than requiring a user to have prior programming knowledge, the web/mobile app makes use of a simple GUI that is capable of creating complex if-this-then-that logic between any connected devices irrespective of their supported protocol.

In a modern world, communication with the outside world is crucial to everything. The infrastructure to support modern communication technologies is often destroyed during natural disasters which complicates the timely search and rescue of victims and the recovery of WIFI infrastructure. This project uses an ad-hoc network of small packages to provide WIFI and are quickly and easily deployable even with damaged transportation infrastructure. This ad-hoc distributed network restores communication to expedite search and rescue and enables rapid recovery of long-term WIFI communication infrastructure.

This project aims to design a secure, decentralized system to track and view the vaccine status of individuals, preventing false vaccination claims. For each vaccine dose administered by a clinic, an non-fungible token is created and transferred to the patient. A person’s vaccination status can then be validated by checking to see if they own a token from a specified clinic list. This solution ultimately digitizes immunization records, providing a tamper-proof, single source of truth solution.

Repeated concussions can have lasting impacts on the cognitive ability for athletes. The goal of this project is to provide a mixed hardware and software-based solution to aid in the detection and prevention of concussions. To accomplish this the SmartHelmet is fitted with a strategically placed accelerometer embedded in the helmet that can detect if an athlete's head meets a threshold force indicating a suspected concussion. This device hopes to fill the market gap in affordable and accurate concussion detection devices available to amateur athletes.

In 2026, the 3D rendering market is expected to be worth 9 billion dollars as more developers and creators are looking to enter the space. QuickScan aims to improve the experience of using photogrammetry by introducing a mobile application with a user guidance interface and machine learning algorithms built-in. These features are aimed at increasing the speed and accuracy of the photogrammetry process with the end goal of making it more viable for the average creator.