Seyyed Moein Kazemi

I am an AI engineer and researcher finishing my Master of Science degree in Data Science at Shahid Beheshti University (GPA: 18.29/20). Prior to this, I completed my Bachelor of Science in Computer Engineering at Iran University of Science and Technology (IUST). Throughout my academic journey, I have achieved elite national rankings in the highly competitive Iranian University Entrance Exams, placing in the top 0.6% (among >160,000 applicants) for my B.Sc. admission and the top 1.5% (among ~15,000 applicants) for my M.Sc. admission.

I am actively seeking fully-funded Ph.D. positions in Computer Science, Artificial Intelligence, and Data Science starting Fall 2027. Please find my detailed Curriculum Vitae here

Research Focus

My primary research interests span the intersections of:

• Efficient Deep Learning & Neural Architectures: Designing ultra-lightweight, parameter-efficient architectures—such as CRNN-KAN, RL-WavKAN, and SW-KAN—for high-stakes and edge computing environments.
• Medical AI & Healthcare: Formulating robust deep learning and computer vision frameworks for biomedical signals and multimodal diagnostic data (e.g., phonocardiogram classification, bone fracture segmentation, and chest X-ray vision-language reporting).
• Deep Reinforcement Learning & Optimization: Exploring various RL methodologies—from value-based approaches to policy optimization—to address different challenges. In my recent work, I have applied these techniques to train custom agents for dynamic neural parameter tuning and to help develop scalable pipelines for Big Data.
• Time Series Forecasting: Modeling dynamic cross-asset correlations, non-Gaussian heavy-tailed distributions, and extreme tail risk in complex temporal sequences via customized likelihood estimators.

Academic & Research Background

At Shahid Beheshti University, I conduct research as a Graduate Research Assistant under the supervision of Dr. Mahboobeh Riahi-Madvar and Dr. Ehsan Bahrami Samani. My recent efforts focus on:

1. Lightweight Medical Signal Modeling: Proposing parameter-efficient CRNN-KAN and 1D RL-WavKAN models that dramatically compress parameter footprints (up to 100x reduction) while maintaining superior diagnostic accuracy and interpretability.
2. Scalable Big Data Systems: Architecting distributed, unsupervised analytical frameworks tailored for Big Data environments, deployed on Apache Spark and leveraging locality-sensitive hashing (LSH) alongside hierarchical policy gradient filtering.
3. Financial Time-Series: Engineering joint probabilistic LSTM pipelines equipped with dynamic Student-$t$ likelihoods to evaluate risk calibration across volatile commodities.

Beyond algorithmic theory, I maintain a rigorous and hands-on engineering foundation across the software and systems stack. Alongside building applied AI pipelines (such as multimodal medical vision-language models and semantic search engines), my practical experience spans constructing semantic knowledge graphs (integrating Neo4j, OWL, and LLM reasoning agents), developing compiler-level toolchains (such as ANTLR4-based AST parsers for static code analysis), and architecting scalable, containerized backend infrastructures (e.g., FastAPI, Django, Docker). Rooted in my computer engineering background, my technical scope also extends to low-level systems, including OS kernel enhancements (xv6) and digital hardware design (VHDL). Complementing my research, I am passionate about academic mentorship, regularly serving as Head TA and Teaching Assistant for core graduate courses such as Machine Learning, Deep Learning, Big Data, and Advanced Data Mining.

---