Arman Kazemi Email & Phone Number

Austin, TX

Sunnyvale, CA, US

Durham, North Carolina, US

Notre Dame, IN, US

• Proposed an in-memory nearest neighbor search operation with FeFET and Flash multi-bit content-addressable memories that achieve software equivalent accuracies for one/few-shot learning applications, while offering.
• Proposed a multi-bit in-memory accelerator for hyperdimensional computing using FeFET crossbars and multi-bit content-addressable memories that achieves accuracies comparable to high-precision software-based.
• Proposed and implemented a new strategy for mapping neural networks to crossbar arrays using TensorFlow, either reducing hardware utilization by ~50% or saving 2 bits of weight representation, while improving device.
• Designed and modeled a CMOS-FeMFET hybrid synapse for training and inference of neural networks using SPICE and TensorFlow, coordinating research efforts with the EE department and achieving improvements of up to 26%.
• Published papers in top conferences/journals such as DAC, DATE, IEDM, etc and served as a reviewer for DATE, JETC, JXCDC, and Front. Neurosci.

Houston, Texas, US

Accelerated data intensive applications using memristor-based in-memory computing.

Shemiran, Tehran, IR

• Designed a serial adder using carbon nanotube field-effect transistors for quaternary logic using SPICE, achieving 42.3% lower power consumption and resulting in a conference publication at ISMVL-2018.

• Trained 5 students in one-on-one classes for the TOEFL and GRE tests, articulating the structures of the exams and empowering them to achieve scores between 90-110 (out of 120) on the TOEFL and 290-310 (out of 340) on.
• Presented free discussion classes aiming to improve the verbal language proficiency of the participants, handling classes of up to 10 people in size and facilitating the improvement of the participants.

Doctor Of Philosophy - Phd, Computer Science And Engineering

Bachelor Of Science (Bs), Computer Engineering - Hardware