About Me
Parsa Abbasi recently completed his MSc in Artificial Intelligence at Iran University of Science and Technology. His thesis focused on enhancing graph attention networks for relational graphs. Specifically, he proposed and developed an improved model called RAGATv2 that adopts a dynamic attention mechanism to better capture relational information while reducing training parameters to prevent overparameterization. Evaluation on standard benchmarks demonstrated superior performance over prior methods.
Parsa received his BSc in Computer Engineering from the University of Guilan, where he explored areas such as natural language processing and deep learning. During his undergraduate studies, Parsa co-authored a research paper on designing sentiment analysis models using deep learning architectures for low-resource Persian language.
Driven by his passion for the field, Parsa seeks to pursue a PhD in AI to continue his research and further advance the field.
- Graph Neural Networks
- Relational Graphs
- Natural Language Processing
- Deep Learning
MSc in Artificial Intelligence, 2020 - Ongoing
Iran University of Science and Technology
BSc in Computer Engineering, 2015 - 2019
University of Guilan
Skills
python (proficient), java (experienced), matlab and c++ (amateur)
keras, pytorch, scikit-learn, imbalanced-learn, h2o
numpy, pandas, plotly, matploblib, sql
Experience
- Designed a comprehensive syllabus covering data science fundamentals for a 60-student bootcamp
- Managed a team of 6 mentors to develop coding assignments, projects, and provide student mentoring.
- Oversaw all educational aspects of the bootcamp, including curriculum planning, instructor management, and the evaluation process.
- Conducting an online Machine Learning course, which included lectures, coding assignments, and a beta-testing process consisting of guiding and monitoring a team of more than 10 beta-testers.
- Led the update and debug procedures for the Data Analysis course, working with a team of three debuggers to enhance the course material and fix bugs.
- Guided students in both the Data Analysis and Machine Learning courses, answering more than 450 clarifying questions and receiving excellent feedback from students (4.9/5.0 average rating).
- Gained expertise in designing educational materials, developing course content and objectives, mentoring learners, and optimizing course feedback mechanisms, all while working with diverse teams.
- Researched Abstractive Text Summarization using cutting-edge deep learning architectures, specifically transformer-based models.
- Deployed a highly successful model that achieved exceptional Rouge scores in generating Persian news headlines.
- Presented a Sentiment Analysis system for Persian language, utilizing deep learning and machine learning models such as Bi-LSTM, CNN, and SVM, and incorporating NLP concepts such as word2vec and fasttext.
- Designed innovative text data augmentation techniques to enhance the performance of sentence-level binary classification and multiclass classification, yielding accuracy rates up to 91% and 67%, respectively.
- Co-authored a research paper which was presented at the Fifth National Conference on Computational Linguistics, Linguistics Society of Iran.
- Implemented an effective Distributed System based on the MPI model, providing parallel computing for computer engineering students.
- Demonstrated proficiency across various tools and methodologies, including Networking, SSH, NFS, and MPICH, among other relevant technologies.
Recent Publications
Teaching Experience
- Successfully instructed a group of 60 students in Natural Language Processing fundamentals
- Utilized popular NLP libraries such as NLTK, SpaCy, Gensim, and HuggingFace, guiding students in hands-on development of real-world NLP tasks
- Delivered engaging instruction on data wrangling and visualization using essential Python libraries, including NumPy, Pandas, Matplotlib, Seaborn, and Plotly to learners of various levels. Demonstrated the practical applications of each tool by providing coding examples based on real-world scenarios.
- Also provided instruction on web scraping concepts and tools, to impart knowledge about data acquisition from web pages.
- Instructed on various machine learning algorithms that are helpful for data analysis tasks and more interpretable, such as decision tree and K-NN algorithms
- Taught web scraping concepts and tools, including BeautifulSoup, Scrapy, and Selenium, to a diverse group of students, providing coding demonstrations and examples from real-world scenarios.
- Successfully led and coordinated a team of five students through exercises and projects, including developing an agile plan to ensure timely completion of their work.
- Evaluated team projects by assessing presentations, critiquing proposed solutions, implementation strategies, and final outcomes, while providing constructive feedback and assigning scores based on objective criteria.
- Collaborated closely with other mentors and organizers and actively participated in weekly discussions to evaluate course objectives and improve student engagement.
- Conducted an engaging workshop on deep learning techniques for image processing, leveraging Convolutional Neural Networks (CNNs), ResNet, and U-Net architecture.
- Developed and identified challenging hands-on projects to provide participants with practical, real-world experience in solving complex image processing tasks, such as Image Classification and Image Segmentation.
- Provided coding projects and homework to supplement educational materials.
- Instructor: Javad PourMostafa Roshan Sharami
- Provided educational materials, including weekly homework assignments.
- Conducted weekly clarification sessions to ensure students understood course material.
- Instructor: Dr. Seyed Mohammadhossein Shekarian
- Collaborated with a team of teaching assistants to provide guidance and assessment for student projects.
- Instructor: Dr. Seyed Abolghasem Mirroshandel
Recent & Upcoming Presentations
Accomplishments
Throughout his academic journey, he has completed various online courses. However, in recent years, he has been relying on several learning resources, such as Youtube videos, online articles, academic papers, and books, to acquire knowledge.
Recent Posts
Contact
- parsa.abbasi1996@gmail.com
- Rasht, Guilan
- DM Me
- We can have a talk on Discord