cv

Master of Science in Artificial Intelligence

• Cumulative GPA: 4/4 or 17.61/20.0
• Thesis title: An approach for improving the robustness of deep neural network image classifiers against adversarial examples

Bachelor of Computer Engineering

• Cumulative GPA: 3.17/4 or 15.90/20.0

Data Scientist

• Working as a data scientist in the R&D team.
• Working on different projects for loan default prediction.

Research Assistant

• Under the supervision of Prof. Maryam Amirmazlaghani
• Research on improving the robustness of Artificial Neural Networks against adversarial attacks

Machine Learning Engineer

• Research & Design AI Products
• Worked on biometrics projects
  • face anti-spoofing and liveness detection using deep learning techniques
  • blink detection and blink counter models using semi-supervised approaches

• 2nd place in The National fNIRS Data Analysis Competition focusing on brain-computer interface systems to help patients with movement disorders.
• National Brain Mapping Laboratory, Tehran, Iran

• Ranked 61st Among More Than 15,000 Participants in the Iranian University Entrance Exam For Masters In Artificial Intelligence.
• Iran National Organization of Educational Testing, Tehran, Iran

• Ranked 36th Among More Than 15,000 Participants in the Iranian University Entrance Exam For Masters In Algorithms and Computation.
• Iran National Organization of Educational Testing, Tehran, Iran

• Ranked 22nd Place in National Collegiate Scientific Olympiad in Computer Engineering.
• Iran National Organization of Educational Testing, Tehran, Iran

• 1st Place in the Provincial Programming Contest of Hamedan
• Hamedan University of Technology, Hamedan, Iran

• selected as a member of Bu-Ali Sina University’s team for the West Asia Regional ACM-ICPC Contest.
• Bu-Ali Sina University, Hamedan, Iran

• 2nd Place in the 2019 Programming Contest, Iran West Region (WICPC).
• Bu-Ali Sina University, Hamedan, Iran

• Ranked 14th in the West Asia Regional ACM-ICPC Contest.
• sharif university of technology, Tehran, Iran

• 1st Place in the 2017 Programming Contest, Iran West Region (WICPC).
• Bu-Ali Sina University, Hamedan, Iran

• Top 2 % Nationwide entrance exam of Iranian Universities, very competitive with nearly 163,000 participants.
• Iran National Organization of Educational Testing, Tehran, Iran

Robust Supervised contrastive learning

• Independently executed the implementation of adversarial training coupled with supervised contrastive learning loss using PyTorch. I thoroughly explored diverse approaches within the PyTorch framework, experimenting to enhance the model’s performance and resilience against adversarial attacks.

multivariate HMM

• Utilizing my programming abilities, I implemented a range of algorithms for Hidden Markov Models (HHMs) in the multivariate case from scratch, leveraging the capabilities of NumPy. These algorithms encompassed alpha recursion, beta recursion, expectation maximization, and viterbi. Through the application of these algorithms, the project aimed to efficiently model and analyze sequential data with hidden states.

PSO-SGD Hybrid Optimizer

• Developed a custom PyTorch optimizer by designing a novel optimization algorithm that combines the strengths of Stochastic Gradient Descent (SGD) and Particle Swarm Optimization (PSO).

Typicality-Based Collaborative Filtering Recommendation

• Implemented a collaborative filtering recommendation algorithm based on the paper ”Typicality-Based Collaborative Filtering Recommendation”. Leveraged item typicality and user preference data to generate personalized recommendations. Tested the algorithm on movieLens-latest-small and TagGenome datasets.

Kohonen Self Organizing Map

• Implemented Kohonen Self-Organizing Map (SOM) neural network from scratch utilizing NumPy. Implemented key algorithms including winner-takes-all, neighborhood function, and weight update. Designed a framework to facilitate training, visualization, and cluster analysis. Leveraged SOM for unsupervised dimensionality reduction on the YaleFaces dataset. Clustered and visualized high-dimensional face data in 2D space.