Resume

• 2017

  Daisy Intelligence Corporation

  University of Toronto

  IBM Canada Ltd.

  McGill University

  IBM

  Toronto

  Canada Area

  Lead Engineer

  Cloud Innovation lab

  IBM

  Toronto

  Canada Area

  Lead Engineer (IBM Canada Ltd.

  Hardware Acceleration Lab

  Toronto) \n\n\tDesign and development of hardware accelerated systems \n\tIntegration of GPUs to IBM DB2-BLU database\n\tFast query processing for costly database operations (Groupby

  join

  sort) on Power8 processors and Nvidia GPUs \n\tHybrid software/hardware algorithms to increase performance on heterogeneous systems (GPU+CPU)\n\n\nIBM Liquid Metal project\n\tDesign and implementation of high performance drivers for liquid metal project\n\tDesign and implementation of fast I/O library for running lime Programs(java based language) on Corsa4 and corsa5 FPGAs\n\tI/O and drivers give lime programs the ability to run synchronously and asynchronously on FPGA\n\tUsing POSIX multi-threading for high speed Asynchronous communication with FPGA\n\nOpenCL Evaluation\n\tPerformance evaluation of Altera’s OpenCL compiler\n\tComparing the performance of OpenCL kernels on GPU

  FPGA and multi-cores

  Engineer at IBM Hardware Acceleration Lab

  IBM Canada Ltd.

  Toronto

  Canada Area

  Parallel CAD\nTransactional Memories\nMulti-cores\nParallel Game Servers\nInstructor of operating Systems at UOT/Scarborough

  Postdoctoral Researcher & Sessional Lecturer

  University of Toronto

  Toronto

  Canada Area

  Sessional Lecturer

  University of Toronto

  . Project lead and architect - VXTW Project at McGill University\n· Designed and implemented a parallel simulator for digital logic simulation.\n · Designed and implemented load balancing algorithms to balance the load during the simulation.\n · Applied Reinforcement Learning techniques to improve the performance of the load balancing.\n · Implemented an optimistic concurrency control between processors.\n · The system was implemented by C++.\n · Message Passing Interface (MPI) utilized for the communication between the processors.

  McGill University

  Toronto

  Canada Area

  Director

  Machine Learning and Software Development

  Daisy Intelligence Corporation

• 2007

  PHD

  Thesis: "Applying parallel Processing and Learning Techniques to Optimize Logical Simulation"

  Computer Science

  Parallel Processing

  \nSimulation and Modelling\nTesting verification\nLeaning

• 2004

  Master of Science

  Thesis: "Routing Algorithms for Necklace-Hypercube Interconnection Network"

  Computer Engineering

  Computer Networks\nSimulation\nHigh Performance Computing

• 2000

  Bachelor of Science

  Computer Engineering; Hardware Engineering

  AmirKabir University of Technology

  Bachelor of Science

  Industrial Engineering; Project Management

  AmirKabir University of Technology

• Machine Learning

  C++

  Software Engineering

  Computer Science

  Linux

  Java

  Programming

  Research

  Operating Systems

  LaTeX

  Distributed Systems

  Algorithms

  C

  Software Development

  High Performance Computing

  C#

  Simulations

  Matlab

  Python

  Optimization

  XMulator: A Listener-Based Integrated Simulation Platform for Interconnection Networks

  Hamid Sarbazi-azad

  A flexible

  easy to extend

  fully object-oriented

  and multilayered simulator for interconnection networks can provide researchers with a great assistant. It is so desirable to attach newly designed components to the current models and to exploit detailed results.\nThis paper presents Xmulator

  an object–oriented listener-based simulation environment for evaluating multicomputer networks. The simulator involves a toolbox of various network topologies

  routing-switching algorithms

  and flexible router models. These router models can vary from complex theoretical architectures to simple real devices. This work introduces a simulator based on listener-based integration which has a great impact on extensibility of the system. Mixed-mode event processing

  improves the performance of the simulator. By decoupling individual parts of the code

  Xmulator enables independent code development and creates a flexible and extensible environment for different aspects of network design. This simulator extensively uses XML format for defining topologies

  parameters

  and outputs which leads to more level of flexibility. To the best of the author’s knowledge that is the first simulator that is able to simulate any arbitrary topology in presence of faults.

  XMulator: A Listener-Based Integrated Simulation Platform for Interconnection Networks

  Sina

  Meraji

  University of Toronto