PhD in Hardware for Machine Learning

The Adaptive Devices and Microsystems lab at the George Washington University conducts research in novel types of devices and hardware foundations for the next generation computing systems. The motivation behind this work is that training of deep neural networks is expected to consume massive amounts of computing resources in the coming decades. Current solutions such as GPUs and TPUs are limited by computing-memory bottlenecks since they can only store a relatively small part of the network model at once. Our research is currently focused on two-terminal non-volatile memory devices called memristors. These devices have shown an electrical behavior similar to that of artificial synapses and can be easily integrated into dense three-dimensional matrices that can store synaptic weight values for neural network models. Find out more at http://adam.seas.gwu.edu.

Come work in our multi-disciplinary group on the development of machine learning training accelerators for neuro-inspired computing using emerging technologies. You will gain a broad range of theoretical and experimental skills related hardware instantiation of digital and mixed signal algorithms for AI/machine learning. The research will be focused novel algorithm development for FPGA / ASIC, quantization aware models, experimental development and testing of hybrid FPGA / nanodevice hardware, etc. Our team collaborates closely with U.S. national labs in the D.C. area, large manufacturing companies and with groups at U.S. and European institutions. The selected student will be mentored to apply for internships and fellowships.

This is a fully funded Ph.D. opportunity in Computer Engineering at George Washington University located in downtown Washington, D.C.! Tuition, salary and health insurance covered. Yearly renewal of funding is contingent upon satisfactory performance.