| Follow @TJzafar
Zafar Takhirov

Zafar Takhirov

HW/SW Engineer

Neighbor Next Door, Santa Clara, CA 95054
(617) 826-9666
English, Russian, Tajik, Turkish, German,



While my work during my schooling focused on digital design, error mitigation, and machine learning, my non-work interests range widely from information theory (go Shannon!), quantum computing, grandfather paradox, Star Trek, Little Mermaid, "why is the grass green?", 1Q84, etc., etc., etc. If you want to talk about, well, anything - just ping me.

Work Experience

Work Experience (6)

  • Research ScientistFacebook Inc.

    Dec, 2018 — Present 1 year 8 months

    Open-Sourcing some AI Stuff

    • PyTorch Quantization

    • Kernel Development

    • Coffee Drinking

    • Research stuff...

  • Software Developer IIDidi Research America, LLC

    May, 2017 — Nov, 2018 1 year 6 months

    Making self-driving cars possible

    • I fooled you, I fooled you, I got pig iron, I got pig iron, I got all pig iron...

    • What can be cooler than robots and lazers? I get to work with both :)

    • Driving using cameras only? Don't think so! Oh wait...

  • Research AssistantBoston University

    Sep, 2011 — Aug, 2017 5 years 11 months

    Worked on several projects as part of ICSG group, including

    • Energy-efficient Machine Learning Algorithms and implementation

    • Error avoidance/mitigation in digital circuits

    • Equalization in chip-to-chip interconnects

  • CTOAvicennas Group, Inc.

    May, 2015 — Feb, 2016 8 months

    Bringing the digital into the healthcare :)

    • NodeJS/AngularJS/ExpressJS/MySQL - Which one doesn't belong? :)

    • Managing a team of 5 engineers

    • Being on coffee 80 hours a week!

  • Mixed-Signal Verification EngineerAnalog Devices, Inc.

    Feb, 2014 — Feb, 2015 1 year

    Worked as part of the automotive group

    • Was responsible for design verification of analog and digital blocks

    • Worked on 3 projects to develop and deploy UVM and real-number modeling environment

    • Participated in functional safety discussion (ISO 26262)

  • Mixed-Signal Design Engineering InternAnalog Devices, Inc.

    May, 2013 — Dec, 2013 7 months

    • Implemented RNM system for simulation speed up (1000x in average)

    • Was involved in mixed-signal design flow

    • Co-developed calibration mechanism for a mixed-signal system


Education (3)

  • Computer Engineering, Ph.D,  Boston University

    2012-09-01 — 2018-01-25

    Dissertation: Designing Energy-Efficient Computing Systems using Equalization and Machine Learning Learning From Data (Machine Learning Algorithms) Artificial Intelligence
  • Computer Engineering, M.S.,  Boston University

    2009-09-01 — 2012-05-24

    Advanced Digital Design Computer Architecture Design of Secure Computer Systems
  • German Linguistics, Specialist (with Distinction),  Russian-Tajik (Slavic) University

    2003-09-01 — 2008-05-24

    Thesis: Neologisms and Slang in German Languages


  • Programming Languages
    C/C++ Python (System-)Verilog HDL MATLAB Scala (Chisel) LaTeX
  • FPGA/ASIC/Hardware Tools
    Power/Timing/Noise analysis Formal verification Verilog-to-Silicon Cadence RTL Compiler Cadence SoC Encounter Icarus Verilog

Projects (7)

  • PyTorch Tutorial Library

    in progress (as of Aug, 2020)

    Python Tutorials for machine learning

  • RISC-V HW Machine Learning Accelerator

    completed (as of Nov, 2017)

    Hardware co-processor for machine learning acceleration written in Chisel (Scala) and designed layout using Cadence SoC toolchain

  • Identification and Authorization using Fingerprints

    completed (as of Mar, 2017)

    Python library for fingerprint identification using Bozorth 3 algorithm to be used as a substitute for username identification

  • Fingerprint Visualization

    completed (as of Apr, 2016)

    Python library for fingerprint visualization. Visualization is done for the ridge flow map of the fingerprints for easier 'manual' recognition

  • Quadcopter with Machine Learning Navigation

    completed (as of Jan, 2015)

    Quadcopter automation tools using FPGAs written in C++/Verilog

  • Background Subtraction using FPGA

    completed (as of Apr, 2014)

    Image background subtraction written in MATLAB and Verilog. The system is implemented on an FPGA and uses a live camera feed to detect anomalies using background subtraction

  • Handwriting Calculator using Neural Networks

    completed (as of Apr, 2014)

    Handwriting recognition system using Multi-Layer Perceptron neural network and implemented in Verilog. The system is trained using Python and tested on Xilinx FPGA board.

Publications & Patents

Publications & Patents (1)



  • Wildlife

    Cats Unicorns/Pixies/Fairies
  • Sport

    Soccer Kickboxing Snowboarding
  • Professional

    Machine Learning High Performance Computing Energy-Efficient Computing