• Full Name:Rezā Latifi
  • Email:reza.latifi[@]mail.um.ac.ir
  • Logo:

Hello There!

I’m currently working as a Computer engineer at Mana Andisheh, which is a subsidiary company of Green Web Holding for more than 2 years.

Prior to my current position, I have been in the IT industry for about 2 years as a programming teacher at a private institude, teaching assistant at the Ferdowsi University of Mashhad and freelance web developer.

I have studied Computer engineering at the Ferdowsi University of Mashhad from 2016 to 2021. I am passionate about software engineering, algorithms, natural language processing and machine learning.

Most of my projects can be found on GitHub. If you’re interested to know more about my experiences, checkout my LinkedIn profile.


  • Experience

  • Project Manager, Full-time

    GreenWeb August 2021–Present

    Using facial recognition technology to track employees entering or leaving a work site could be the solution employers are looking for. As the first step to achieving this goal, authenticating the arrival and departure of the employees using their phone front camera can be helpful. The aim of this project is to use face recognition for authentication, which can be used in various use cases. Further steps are the ability to recognize gestures and poses. The model trained in the project can be reused using the Transfer-learning methods as a general face descriptor.

  • Software Engineer, Full-time

    GreenWeb June 2021–August 2021

    Developed an API for managing Site-builder affiliates credit in WHMCS. The API is frequently called. Therefore, the challenge was the response speed and reliability of the service.

    Another project I developed is the Site-builder affiliate software as an addon in WHMCS, which is a complete toolkit for Site-builder affiliates to set up their own Site-builder service.

  • Back End Developer, Part-time

    GreenWeb September 2019–June 2021

    Developed a software for Site-builder affiliates in WordPress. Mana Site-builder creates websites within seconds, providing hundreds of professional website templates and tens of easy-to-use management tools. The affiliate software is responsible for using the Site-builder API to create websites. Providing management tools for both affiliates and customers, and the ability to sell services using Woocommerce online shop products. I was responsible for software architecture, database design, and backend programming with PHP.

    Developed an accounting service for customers of the Site-builder to connect their online shops to some common accounting softwares.

  • Tutor, Full-time

    Private institude June 2019–August 2019

    Tutored PHP and MySQL to a class of 10-20 Pupils. The final project of the course was a Telegram English-Persian translator bot.

  • Teaching Assistant, Full-time

    Ferdowsi University of Mashhad February 2019–June 2019

    Professor: Dr. Mahmoud Naghibzadeh

    As a TA of “Algorithms Design” undergrad course I was responsible for testing and grading of assignments, and providing students with informative feedback. Moreover, I did hold extra training classes for students who asked for.

  • Research & Development, Full-time

    FUM Robotics Community 2016–2021

    Developed an admin panel for customer management of PasokhPlus with Laravel and PHP.

    Developed a Joomla plugin to publish, delete or update articles of a news agency website in a specified Telegram channel automatically with PHP.

    Developed a Telegram bot for receiving the news for a news agency website. Providing management options for managers.

  • Education

  • B.Sc. in Computer Engineering

    Ferdowsi University of Mashhad 2016–2021

    Graduated with a cumulative Overal GPA of 3.04/4.00 and Major GPA of 3.24/4.00
    and a degree project entitled
    “Refining Fārsi(Persian) text from loan words using machine learning”.

  • High School Diploma (Mathematical and technical)

    National Organization for Development of Exceptional Talents (Sampād) 2012–2016

    Activities: Library management assistant


Bachelor degree project


Arnavāz [Who is Arnavāz?] refines Fārsi(Persian) text from destructive loan words. The model is based on GPT-2 pre-trained model and fine-tuned using purified rewrites of Fārsi books alongside rewrote sentences of Mozilla CommonVoice by volunteers.

The public corpus of Arnavāz is the first and only collection of its type in Fārsi. It is free to use.

Offline map matching on the output of GPS-free car location system

Map Matcher

The map graph is created from a network extracted by SUMO from OSM. The calibration is by using the latitude and longitude of the starting point. The passed distance and heading must be given at each step.

The applied approach is finding the nearest location to the last recorded location from the candidate locations using the Dijkstra algorithm.

Since the project is part of a PhD thesis. I can not share the source code.
Image descriptor for human detection

Histograms of oriented gradients for human detection

Implementation of HOG based on the paper by Dalal and Triggs. Furthermore, I have prepared a program to find gradients using Maclaurin-Series. The program can be applied instead of Sobel for finding gradients in HOG.

Small, fast, lightwight WordPress plugins

WordPress Micro Plugin Boilerplate

A standardized, organized, object-oriented, mimimal foundation for building high-quality Services or tiny WordPress plugins.

Final project of Advanced Programming undergrad course

ٰVoting machine

The project consists of two separate programs, the Server and the Client. The Server program looks for a free port starting from 8000 and opens a socket at the first available port. The candidates will be passed to the Clients when they connect to the Server. The votes and authentication credentials will be sent to the Server through the Socket.

Final project of Information Retrival undergrad course

Search Engine

A search engine for advertisements scraped from Divar.

Using sub-words for indexing and search queries elevated the performance of the traditional unigram language model with smoothing.


Published in 11th International Conference on Computer and Knowledge Engineering (ICCKE2021)

A New Hypercube Variant: Pruned Shuffle Connected Cube

The performance of System-on-Chip (SoC) strongly depends on the architecture of interconnections. Hypercube, an extended analogy of a three dimensional cube which covers higher than three dimensions, is a popular and well-known interconnection network. In this paper, we propose a variant of the hypercube named Pruned Shuffle Connected Cube (PSCQ). PSCQ maintains all desired properties of the hypercube with reduced node edges, lower diameter length, lower average distance, and lower average vertex degree. Furthermore, some limitations of the hypercube are obviated. Knowing that minor improvement in these measures can have a great impact on the efficiency of the model, many improved versions of hypercube have been proposed. However, the proposed version of this research improves all performance measures. The detailed architecture and comparing of topological parameters are discussed and comparative results with state of the art versions are presented.


Have an interesting Computer engineering/Science research idea?

Want to hire an experienced Computer engineer?