Ali J. Alaee

In 2021, I joined the MLSys Lab as a research assistant under the supervision of Prof. Hamid R. Rabiee and Dr. Abbas Hoseini. My primary involvement with this team has been in my thesis project, developing an effective retrieval method based on the content of queries and items and the history of users' behavior. Given a query as an input, retrieving and ranking items is critically important in e-commerce platforms. My project aims to propose and develop an effective personalized embedding-based ranking algorithm, to improve users' experience. The main idea is from Word2Vec, an NLP algorithm that has been broadened to Search2Vec for utilization in recommendation systems. A trained neural network based on the mentioned algorithms produces an embedding space that can effectively place related queries and items near each other.
Moreover, user clustering, based on users' history of clicks, has been employed to personalize the results of our ranking method. Two datasets, one obtained from torob.ir (one of the largest Persian online shops), and the other from AOL (publicly available on Kaggle), are being used for experimenting. Implementation of this project with TensorFlow plus weekly progress reports are available in this github repository.

I started my contribution to the private meta-learning project in a group of three in September 2021 by joining the meta-learning group. In this research project which Prof. Hamid R. Rabiee is the supervisor, We are investigating the threats of reconstruction attacks on meta-learning models. Furthermore, we are proposing a proper defense mechanism for addressing this vulnerability.

In July 2019, I joined a start-up called Pido, One of the largest Fuel Delivery companies in the Middle East. I mainly worked on a project which helped advertisers target relevant users to get the most revenue. In this project, we developed a recommender system based on users' and products' metadata and the history of purchases. Our approach was to classify users and products and match them based on their semantic embeddings. An offline implementation of this project based on our method is now deployed in the company's application. Moreover, as a minor project, I have developed a feedback analyzer method that extracts insights from users' reviews corresponding to each service provider.

After I finished my career in the chemistry olympiad as a student, I started my job at Helli-3 high school. I designed a professional curriculum with the help of experienced teachers, organized exams and evaluations, and supervised talented students (5 - 10 students per year) to prepare them for the olympiad competition. As the mentor and academic advisor of the Chemistry Olympiad team, I have helped my students achieve outstanding results. At IChO 2O21 in Japan (53^rd International Chemistry Olympiad), Mahbod Alian, a team member, won an international gold medal. In addition, he and one of the other students of the department have won national gold and bronze medals, respectively.

Some of my teaching experiences as an assistant are:

• Artificial Intelligence - Prof. Rohban (Fall 2021)
• Artificial Intelligence - Prof. Rohban (Spring 2021)
• Linear Algebra - Prof. Abolfazl Motahari (Spring 2021)
• Probability and Statistics - Prof. Abolfazl Motahari (Fall 2020)
• Introduction to Bioinformatics - Prof. Sharifi-Zarchi (Fall 2020)

Ranked 7^th in the 26^th National Chemistry Olympiad among over 40,000 participants.

Ranked 27^th in the 25^th National Chemistry Olympiad among over 42,000 participants.