Mohammadreza Karimi

PhD Student

Education

  • Ph.D. Candidate in Artificial Intelligence, Amirkabir University (Current)- 2020 September-Present
    • Statistical Data Analysis Lab
    • Learning Functional Brain Connectivity Analysis from fMRI Data using Graph Neural Network
  • M.S. in Secure Computing, Sharif University - 2017 April- 2021 February
    • Data and Network Security Lab
    • Evaluating the Robustness of Deep Learning-Based Network Traffic Classifiers against Adversarial Attacks
  • B.S. in Information Technology, Sharif University - 2013 September- 2017 Auguest

Experience

  • Data Engineer and Backend Developer, Ministry of Cooperatives, Labor, and Social Welfare (MCLS), Oct 2022 - Present
    • Worked on a Big data and Developing ETL pipline.
    • Developing CRM web services using , its data was fed in a special way into the database using ETL. FastAPI, SQLAlchemy, SQL Server, and Redis
  • CTO and Backend Developer Anya, Oct 2020 - Present
    • BrainGym was an international Iranian marketer’s project. Sequelize-Express-NodeJs.
    • iCritic was a social network for film, music, book, and game reviews. Docker-Postgres-Django.
    • Darsplay was a semi-realtime educational game similar to Quiz of Kings, which was developed for schools. Docker-Postgres-Django.
  • Data Scientist, Mar 2019 - Sep 2019
    • Worked on a project similar to my master’s project, which was a machine learning project for network traffic classification. Xgboost, Catboost, Sklearn.
    • Worked on the reverse of previous project, using adversarial learning to modify network traffic in such a way as to reduce the accuracy of traffic classification by machine learning models. Keras ,TensorFlow, Wireshark, tcpdump.
  • Freelance Programmer, Mar 2019 - Sep 2019
    • Worked on a Telegram crawler project that collected Telegram content by joining hundreds of channels and groups. The ultimate goal of the employer was to create a search engine for Persian Telegram groups and channels

Technical Skills

  • Development: Python, Django, Django Rest Framework, FastAPI, SQLAlchemy, Bash Scripting, JavaScript, Node JS, Express
  • Machine Learning: PyTorch, PyTorch-Gemetric, Keras, Scikit-learn, XGBoost, LightGBM
  • Databases: PostgreSQL, Redis, PyMongo, Elasticsearch
  • Other Skills: Docker, CI/CD, Git, ETL, Nginx, Swagger, Metabase, Kibana, RabbitMQ

Honors

  • Ranked 379th among more than 250,000 participants in the national university entrance exam for B.Sc. degree
  • Ranked 5th among more than 19,000 participants in the national university entrance exam for M.Sc. degree
  • Top-talent Ph.D. acceptance in Computer Engineering Department of Amirkabir University of Technology

Publications

  • Mohammadreza Karimi, Rasoul Jalili, “Enhancing Network Traffic Privacy Against Classifications Attack using Adversarial Learning” (In Persian); ISCISC 2020, Invited Talk. link

Doctorate Thesis

My doctoral research focuses on the analysis of functional connectivity in the brain using functional Magnetic Resonance Imaging (fMRI) data. The inherent graph structure of the brain’s functional connectivity network makes Graph Neural Networks (GNNs) an ideal choice for this task. However, fMRI data is not merely a graph. It is a large multi-variate time series with spatial-temporal dimensions, and intrinsically, it is a 4-dimensional point cloud. This unique nature of fMRI data has been underexplored in the existing literature on learning brain functional connectivity. In my dissertation, I aim to address this research gap by developing a GNN-based model that encapsulates all these aspects of fMRI data. The ultimate objective is to leverage these insights to develop predictive models that can aid in the diagnosis and treatment of diseases.

Contact

  • Personal Email: mokarimi@outlook.com
  • University Email: mo_karimi@at.ac.ir