About
Hi, I'm a final-semester computer science undergraduate student at Amirkabir University of Technology in
Iran.
I work on Large (Vision) Language Models for both research and industrial applications, and
I'm also a full-stack web developer.
I'm only available via email, as I don’t use
any social media!
Research Interests
| ✅ Large (Vision) Language Models | ✅ Multimodal Artificial Intelligence |
| ✅ Synthetic Data Generation | ✅ L(V)LM Evaluation & Benchmark |
Highlights
- • I ranked in the top 0.5% of Iran's nationwide university entrance exam and was admitted to Amirkabir University of Technology as a distinguished student (certificate).
- • I'm on track to graduate early, completing my bachelor's degree in just six semesters, which is a year earlier than the standard timeline, by pursuing the maximum permitted number of courses per semester while maintaining a GPA of 3.92 out of 4.0.
- • Due to early graduation, I had only 2.5 years for research — a period that was very intense given academic pressure, part-time work, an internship, and a daily 4-hour commute to university. My university lacked both suitable GPUs and funding to use proprietary models, so I had to rely on limited public resources like Kaggle. Despite these constraints, I authored four first-authored research papers, including a state-of-the-art long paper in my third undergraduate semester, now under revision for a Q1 journal, and a survey paper on bridging the concepts of VLMs and synthetic data for the first time by reading and analyzing 125 papers and writing the paper — within just 10 days.
Publications
A Survey on Bridging VLMs and Synthetic Data
Mohammad Ghiasvand Mohammadkhani, Saeedeh Momtazi, Hamid Beigy
Under Review at Multimedia Tools and Applications
Gap-Filling Prompting Enhances Code-Assisted Mathematical Reasoning
Mohammad Ghiasvand Mohammadkhani
arXiv preprint arXiv:2411.05407
Zero-Shot Learning and Key Points Are All You Need for Automated Fact-Checking
Mohammad Ghiasvand Mohammadkhani, Ali Ghiasvand Mohammadkhani, Hamid Beigy
FEVER Workshop at EMNLP 2024
E2TP: Element to Tuple Prompting Improves Aspect Sentiment Tuple
Prediction
Mohammad Ghiasvand Mohammadkhani, Niloofar Ranjbar, Saeedeh Momtazi
2nd-Round Review at Neural Networks (Q1) Journal — Revision Submitted
* Completed in my 3rd undergraduate semester *
Industrial Experience
- • Teamyar ERP Company: I completed a one-month AI summer internship (certificate) at this company, then worked as a freelancer on a project focused on retrieval-augmented generation, where I developed a pipeline for customer service that answered customer questions by grounding the LLM in the company’s extensive software documentation.
- • Freelance Web Developer: I built the backend using Django and DRF and managed frontend–server communication with Vue.js for web applications on a contract basis.
Professional Services
- • Conference Workshop Reviewer: MATH-AI at NeurIPS 2024, FEVER at EMNLP 2024
Language Proficiency
- • English: (IELTS) 8.0/9.0 (Listening, 8.5 | Reading, 8.0 | Speaking, 7.5 | Writing, 7.0)
- • Persian: Native