Shreyas Donti

• Designing custom PCBs in Altium Designer for a wearable wind-powered sensing system, optimizing for low power, compact form factor, and sensor integration. • Leading end-to-end design and firmware implementation of a high-precision sensor system on the ADICUP3029 (ARMv5) using C and SPI; acheiving 99.5% impedance accuracy. • Designing an energy-harvesting system for a small wind turbine using a buck-boost controller and LDOs to maintain voltage between 2.2V - 3.5V across circuit • Investigating and evaluating LoRa communication modules to design a long-range wireless system.

• Teach core systems programming concepts including C, x86 assembly, CPU architecture, virtual memory, threading, and networking to 60+ students. • Lead two weekly lab sessions; assist with debugging, systems design, and conceptual understanding. • Grade 50+ assignments weekly and provide feedback to support student learning and performance

• Part of the Onboard Processing hardware sub-team, investigating model deployment on FPGA for hyperspectral image processing. • Working with FINN and Vivado/Vitis to compile PyTorch model to FPGA bitstream for AMD Xilinx Zynq Ultrascale+ MPSoc XCZU2CG development board. • Utilizing Docker and Git to manage development environment and version history respectively.

• Programmed ESP32 to operate electrical sensors and motor control of a ball bearing sorting robot for the ASME SDC 2025 competition, acheiving 3rd place out of 20 national teams, sorting 80 balls within <80 seconds. • Configured stepper motors and LiPo batteries using multimeters, oscilloscopes and other electrical testing equipment. • Designed and soldered custom motor controller PCBs using KiCad for robot actuation

• Built the HackUMass XII & XIII website using React.js, HTML, CSS, and Figma alongside a 10+ member development team, improving user experience and project workflow via Git, and automating deployment with Github Actions. • Delivered performance and UX improvements to website, increasing user registrations by 15% and sponsorship funding by 6%. • Spearheaded the development of a hardware portal utilizing Redis, SvelteKit frontend, and a Go backend hosted on Microsoft Azure to facilitate checkout and return of over 200+ hardware components. • Worked on a judging platform using Node.js, MongoDB, and Express to manage 150+ projects and automate judge assignments and scoring.

• Was elected Webmaster of the UMass IEEE chapter for 2024-2025. • Managed WordPress website, made posts, improved readability by updating style and visuals.

• Organized the Pioneer Valley Model United Nations II and III conferences hosted by UMass, in collaboration with other colleges of the Five College Consortium, increasing registration by 200% in between conferences.

• Worked alongside 2 other students under guidance of PhD mentor to investigate human activity recognition using IMUs. • Designed a 1-D CNN deep learning model using PyTorch and Pandas to classify human activities with 92.5% accuracy. • Employed signal processing techniques on triaxial gyroscope and accelerometer data using scikit-learn and Pandas/NumPy to optimize model; produced +18% accuracy increase in readings. • Presented summary of findings using posterboard with graphs created using Matplotlib and Seaborn to over 100+ students.