Research
Analog Solvers for Real-time Model Predictive Control
September 2019
The College of New Jersey
Advised by Dr. Ambrose Adegbege
- Helped develop MATLAB simulations for comparing different continuous-time quadratic programming solvers
Hardware Acceleration for AI-based Hearing Aids
Ongoing
Illinois Institute of Technology & Northwestern University
Advised by Dr. Kyle Hale (IIT) and Dr. Prem Seetharaman (Northwestern)
- Developing low latency & low power architectures for real-time audio source separation
- Built preliminary RTL designs using Vivado High-Level Synthesis and Chisel HDL
Digital Model Predictive Control (MPC) with FPGAs
June 2018-November 2019
The College of New Jersey
Advised by Dr. Ambrose Adegbege
- Implemented a model predictive control system on an Artix-7 FPGA and deployed it to a Quanser Quadruple-Tank Water system
- Developed custom RTL to accelerate a primal-dual Quadratic Programming solver (2x faster than MATLAB’s quadprog)
Image Generation with Non-Conventional GANs
May 2019-November 2019
The College of New Jersey
Advised by Dr. Larry Pearlstein
- Implemented a custom GAN architecture that utilizes a C-based image generator instead of a differentiable neural network
- Ported existing model implementations from Apache MxNET to Google’s TensorFlow