Deng Pan

EDUCATION

Ph.D. Candidate - Electrical and Computer Engineering
Univeristy of Toronto, 2017 - present, GPA: 4.00/4.00

B.A.Sc in Engineering Science with Honors Distinction
Univeristy of Toronto, 2013 - 2017

AWARD & HONORS

EDWARD S. ROGERS SR. GRADUATE SCHOLARSHIP

DEPARTMENT OF ELECTRICAL AND COMPUTER ENGINEERING, UNIVERSITY OF TORONTO, 2017 - 2018

UNIVERSITY OF TORONTO FELLOWSHIP

SCHOOL OF GRADUATE STUDIES, UNIVERSITY OF TORONTO, 2018

NIKOLA TESLA SCHOLARSHIP

DEPARTMENT OF ELECTRICAL ENGINEERING, COLUMBIA UNIVERSITY, NY, USA, 2018
AWARDED FOR SUPERIOR UNDERGRADUATE ACADEMIC RECORD

DEAN'S HONOR LIST

FACULTY OF APPLIED SCIENCE AND ENGINEERING, UNIVERSITY OF TORONTO, 2013 - 2017

UNIVERSITY OF TORONTO SUMMER RESEARCH EXCHANGE PROGRAM (SREP) AWARD

CENTER FOR INTERNATIONAL EXPERIENCE, UNIVERSITY OF TORONTO, 2016
AWARDED TO 6 STUDENTS AMONG 450 APPLICANTS

ENGINEERING SCIENCE RESEARCH OPPORTUNITIES PROGRAM (ESROP) GRANT

DIVISION OF ENGINEERING SCIENCE, UNIVERSITY OF TORONTO, 2015
AWARDED TO 12 STUDENTS AMONG 385 APPLICANTS

PRESIDENT'S ENTRANCE SCHOLARSHIP

UNIVERSITY OF TORONTO, 2013

WORK AND RESEARCH EXPERIENCE

UNIVERSITY OF TORONTO - RESEARCH ASSISTANT

JUNE 2017 - PRESENT

●   Designing a neural network supercomputer for image-classification and visual recognition used in deep learning
●   Deploying a tree structure multi-tile architecture to accelerate inferencing as an alternative to GPUs and FPGAs
●   Development of a software model of machine-learning accelerator to exploit parallelism in flow of weights and activations
●   Joint-research with Vector Institute and COHESA Research Network for ISA applications

UNIVERSITY OF TORONTO - TEACHING ASSISTANT

JANUARY 2018 - PRESENT

●   Undergraduate Engineering Research Day (UnERD), August 2018
●   ECE302, Probability and Applications, Winter 2017

MONTREAL INSTITUTE FOR LEARNING ALGORITHMS - DEEP LEARNING INTERN

JANUARY - APRIL 2019

●   Designing efficient algorithms for tailored inference architectures using software-hardware co-design paradigm
●   Improving peak performance and energy efficiency in deep learning computing in tandem to preserving model precision
●   Reducing computational complexity of matrix multiplication in neural networks through low rank tensor decomposition
●   Applying iterative pruning techniques to miniaturize memory footprint of fully connected and convolutional layers

HONG KONG UNIVERSITY OF SCIENCE AND TECHNOLOGY - ELECTRICAL ENGINEERING INTERN

JUNE - AUGUST 2016

●   Invented a novel computational model for predicting SRAM voltage margin variation aided with Monte-Carlo simulation
●   Optimized and modeled SRAM circuits using 7nm FinFET technologies for dynamic voltage scalable applications
●   Implemented emerging FinFET-based IC systems and semiconductor devices, focusing on memory circuits
●   Presented at HKUST-SENG Workshop on Semiconductor Challenges and Innovations

NATIONAL UNIVERSITY OF SINGAPORE - SOFT ROBOTICS RESEARCH INTERN

MAY - JULY 2015

●   Researched on soft polymers for robotic artificial muscles and motion-based energy harvesting
●   Improved the Cole-Cole permittivity model for VHB-class dissipative elastomer using dielectric spectrometry results
●   Prototyped acrylic frames in SolidWorks reducing fringing effects from flux leakage in data collection

HOSPITAL FOR SICK CHILDREN (SICKKIDS) - DATA ANALYSIS ASSISTANT

MAY - JULY 2014

●   Measured gene expression area of in-situ hybridization to determine mRA level in heart and placenta samples
●   Analyzed figures using ImageJ and Adobe Photoshop to determine cell counts
●   Documented statistical differences between mutant and wildtype figures in detailed research report

SELECT PROJECTS

OPTIMIZATION IN MOBILE CLOUD AND EDGE COMPUTING

ECE1505, UNIVERSITY OF TORONTO, JANUARY - APRIL 2018

●   Numerical simulation to a resource allocation problem in MCC/MEC systems
●   Devised a non-convex QCQP formulation and solved using SDR techniques
●   Applied a game-theoretic approach by attaining a Nash Equilibrium solution of the joint energy-delay objective

NEURAL NETWORKS FOR VISUAL RECOGNITION

ECE521, UNIVERSITY OF TORONTO, MARCH - APRIL 2018

●   Investigated the classification performance of neural networks on MNIST dataset of 28x28 images with 10 classes
●   Implemented fully-connected and convolutional layers with dropout and L2-regularization to combat overfitting
●   Performed random search algorithms for hyperparameter tuning and weight visualization in Tensorflow and Python

ADVANCED VLSI METHODOLOGIES - SYSTEM DESIGN PROJECT

ECE1388, UNIVERSITY OF TORONTO, SEPTEMBER - DECEMBER 2017

●   Designed an incremental second-order delta-sigma ADC using 0.13um technologies in team of 4 graduate students
●   Improved upon and satisfied system requirements specified in selected ISSCC 2016 reference paper

PRAXIS II RFP RESPONSE CHALLENGE

TEAM LEADER, UNIVERSITY OF TORONTO, JANUARY - APRIL 2014

●   Led a team of 4 students in the redesign of cabinet system for Toronto EMS ambulances
●   Interviewed paramedics to learn design requirements and produced prototype drawn on GoogleSketch
●   Received positive feedback and earned stakeholder approval from paramedics in final design showcase

SKILLS

COURSES

REFERENCES AVAILABLE UPON REQUEST