- Computer Architecture Processor Design, Interconnects, Memristive Crossbars
- Neuromorphic Computing Brain Inspired Architectures, Neural Networks
- Embedded Systems Microcontrollers, RTOS, FPGAs, Sensors
- Autonomous Robotics Small-scale autonomous robots using off the shelf components
- Parallel Computing OpenMP, MPI, CUDA, pthreads, Cluster Design
DANNA - Dynamic Adaptive Neural Network Arrays
Member of UTK's Neuromorphic Computing Team - specifically working with the hardware implementation and hardware-software interaction. Previous tasks include creating an automated build environment, overhauling internal random address selection, improving design timing, and implementing new op codes and features. My thesis is the development of DANNA2 which will serve as the long-term successor to DANNA.
- ARM Supercomputing Working with commodity ARM hardware to develop strategies for High Performance Computing. Examining the benefit of heterogeneous architectures including both varying CPU and GPU architectures working together in a cohesive environment.
- DANNA2: Dynamic Adative Neural Network Arrays International Conference on Neuromorphic Systems 2018 (Accepted, Lead author, presenter)
- NeoN: Neuromorphic Control for Autonomous Robotic Navigation IEEE International Symposium on Robotics and Intelligent Sensors 2017 (Lead author, co-presenter)
Parallel Evolutionary Optimization for Neuromorphic Network Training
Machine Learning in HPC Environments, Supercomputing 2016 (co-author)
- DANNA v2 Working to develop a follow up, next generation neuromorphic architecture as the successor to DANNA (Dynamic Adaptive Neural Network Arrays). I am responsible for the architecture design, RTL implementation, software simulation, network training, and verification.
General Robotics Development Platform
Developing a generalized rapid prototyping platform for small scale autonomous robots. The system will be open source and documented.
IEEE SouthEastCon 2018 Hardware Competition
Served as Team Captain for 2017-2018.
NeoN - A Neuromorphic Navigation System
Serving as team lead for a Senior Design project to develop a collision avoidance system for an autonomous robot using a Neuromorphic architecture (FPGA implementation of DANNA as the Neuromorphic system).
IEEE SouthEastCon 2017 Hardware Competition
Built an autonomous robot to complete a Star Wars themed compeititon. Served as Team Captain for 2016-2017.
- Many Core RISC-V NoC Graduate VLSI Course — Logic and VLSI design of a many-core RISC-V based network on chip processor.
Heterogeneous ARM Supercomputing Cluster
Graduate Supercomputing Course — Building a cluster from 32 Pine64 and 12 nVidia TX1 nodes to form a high performance 64-bit ARMv8 and CUDA heterogeneous cluster. I am focusing on the systems architecture, CUDA integration, and performance evaluation and optimization.
ARM Supercomputing Cluster
Graduate Supercomputing Course — Project based — Building a cluster from 64+ Raspberry Pi 3 boards. I primarly focus on the System and Networking details in addition to optimizing BLAS and HPL for the system.
IEEE SouthEastCon 2016 Hardware Competition
Github Repo Link
Built an autonomous robot to complete a logistics themed task. The robot was required to identify and sort wooden blocks of different sizes and colors.
- Multiplayer Checkers on an FPGA VHDL implementation of multiplayer Checkers on a Xilinx Artix 7 FPGA - including game logic, 4bpc VGA output, joystick support, and move highlighting.
Disk based B+ Tree Implementation
Code available on request
Built a B Tree data structure in C interfacing with an emulated hard drive system. Managed tree structure in an optimal manner with direct sector reads and writes.
- IEEE SouthEastCon 2015 Hardware Competition Built an autonomous robot to complete a series of mini games including playing Simon, drawing IEEE on an Etch-a-Sketch, picking up a playing card, and rotating a Rubik's Cube.
- ProjClock — Multi-user Project Time Tracking Software Built as a responsive web application with a PHP 5 and SQLite backend. ProjClock allows for shared and private project time tracking. It offers both CSV and XLSX (Excel) export options. Completed prior to arriving at UT.