Fateme Mazdarani

I am a second-year PhD student in Computer Science at Clemson University, where I work in the Human-AI Empowerment Lab under the supervision of Dr. Carlos Toxtli. My research broadly focuses on artificial intelligence and machine learning, particularly the reliability, robustness, and interpretability of intelligent systems.

I study large language models and mathematical reasoning, including how models access, evaluate, and verify knowledge across different representations and how retrieval can support more reliable reasoning. My broader interests include human-centered and trustworthy AI, recommender systems, data mining, and scalable machine learning methods for real-world applications.

I am always open to research collaborations and new ideas. Feel free to get in touch or explore my work on GitHub.

How to pronounce my name: FAH-teh-meh (like "fa" in father, "te" in ten, "me" in men).

Selected Publications & Manuscripts

• Published
  Worker-Centered AI: Transparent Explanations for Trustworthy Task Recommendation in Crowd Work.
  Fateme Mazdarani, Alberto Campos Hernandez, and Carlos Toxtli.
  2026 IEEE Conference on Artificial Intelligence (CAI), pp. 295–302.
• Accepted for Publication
  Beyond the Answer Key: Robustness Evaluation of Large Language Models for Step-Level Mathematical Verification.
  Fateme Mazdarani and Carlos Toxtli.
  2026 IEEE International Conference on Machine Learning and Applications (ICMLA).
• Accepted for Publication
  Randomized SVD Approximations for Spectral Co-Clustering of Word-Document Matrices.
  Fateme Mazdarani and Carlos Toxtli.
  2026 IEEE International Conference on Machine Learning and Applications (ICMLA). Code
• Accepted for Publication
  TREAT: Evaluating Access to Formal Knowledge across Equivalent Mathematical Representations.
  Fateme Mazdarani and Carlos Toxtli.
  2026 International Symposium on Symbolic and Numeric Algorithms for Scientific Computing (SYNASC). Dataset & Code
• Accepted for Publication
  The Shape of Mathematical Creativity: Measuring Mathematical Exploration in Formal Proof Generation.
  Fateme Mazdarani and Carlos Toxtli.
  MathAI Workshop at NeurIPS 2026.
• Accepted for Publication
  From Recognition to Reconstruction: Towards Robust Procedural Mathematical Reasoning.
  Fateme Mazdarani and Carlos Toxtli.
  The 2026 Conference on Empirical Methods in Natural Language Processing (EMNLP 2026); Workshop on Mathematical Natural Language Processing.
• Under Review
  The Models Behind the Claim: Empirical Standards for LLM Evaluation.
  Fateme Mazdarani and Carlos Toxtli.
  International Conference on Learning Representations (ICLR).

Research and Teaching

• Graduate Research Assistant, Clemson University  |  Fall 2026–Present
  Human-AI Empowerment Lab, supervised by Dr. Carlos Toxtli. Research in the lab since August 2025.
  Research includes LLM reasoning and evaluation, worker-centered AI, scalable spectral clustering, and simulated autonomous underwater vehicle navigation using vision-language-action models, object detection, and safety checks.
• Graduate Teaching Assistant, Clemson University  |  2025–2026
  Design and Analysis of Algorithms and Human-Computer Interaction / Human-Centered Computing courses.

Academic Service

• Social Chair, School of Computing Graduate Student Association  |  Summer 2026–Present
• Delegate, Clemson Graduate Student Government (GSG)  |  Fall 2026–Present
• Peer Reviewer
  Reviewed 10+ papers at various conferences.
• Poster Judge, Creative Inquiry undergraduate research program, Clemson University
  August 2026.

Selected Presentations & Posters