Amirhossein Dabiriaghdam

Ph.D. Student & Researcher | LLM Agents and World Models | Vancouver, Canada

prof_pic.jpg

Hi there, I am Amir! 👋

I’m a Ph.D. student in Electrical and Computer Engineering at the University of British Columbia (UBC), where I am co-supervised by Prof. Giuseppe Carenini and Prof. Lele Wang. I hold a UBC Four Year Doctoral Fellowship.

I hold a master’s degree from UBC as well, and a B.Sc. degree from the University of Tehran. During my undergrad, I also did an internship at EPFL LTS4 lab, where I worked under Prof. Pascal Frossard on adversarial attacks against neural machine translation models.

My research is on large language models (LLMs), vision-language models, LLM agents, and world models. I develop agentic multimodal systems for reasoning and decision-making in interactive visual environments, with a focus on visual tool use, anticipatory reasoning, and multi-agent collaboration. I also build benchmarks for multimodal reasoning and distributional generation to understand model capabilities, failure modes, and generalization.

Check out my CV and publications for more details.

Short on time? Just ask the LLM what you want to know about me; it knows my papers, background, and what I’m working on.

Selected Publications [view all]

  1. EMNLP 2025
    SimMark: A robust sentence-level similarity-based watermarking algorithm for large language models
    Amirhossein Dabiriaghdam, and Lele Wang
    EMNLP Main, 2025 Oral (325/8174 = 3.98%)

    Introduces a black-box sentence-level watermark based on semantic similarity that remains detectable after paraphrasing while preserving text quality.

  1. arXiv
    VAMPS: Visual-Assisted Mathematical Problem Solving Benchmark
    Amirhossein Dabiriaghdam, Shayan Vassef, Mohammadreza Bakhtiari, Yasamin Medghalchi, Ilker Hacihaliloglu, Mesrob Ohannessian, Lele Wang, and Giuseppe Carenini
    arXiv preprint arXiv:2606.04244, 2026

    Introduces a bilingual multimodal benchmark for studying whether vision-language models can use graphs and visual tools to solve mathematical problems.

  1. arXiv
    UnpredictaBench: A Benchmark for Evaluating Distributional Randomness in LLMs
    Amirhossein Abaskohi*, Amirhossein Dabiriaghdam*, Liang Luo, Ellie Dingqiao Wen, Lele Wang, Giuseppe Carenini, and Peter West (* Equal contribution)
    arXiv preprint arXiv:2606.06622, 2026

    Introduces a benchmark and KS@N metric for testing whether language models reproduce target probability distributions when sampled repeatedly.

  1. ICASSP 2023
    Targeted Adversarial Attacks Against Neural Machine Translation
    Sahar Sadrizadeh, Amirhossein Dabiriaghdam, Ljiljana Dolamic, and Pascal Frossard
    ICASSP, 2023

    Introduces a targeted white-box attack that inserts chosen keywords into neural machine translations while preserving source-sentence similarity.

Teaching View CV

AI 100: Introduction to Artificial Intelligence

Graduate Academic Assistant · UBC · June–September 2026

I helped develop a broad-audience introductory AI course, contributing to:

  • Course content and instructional materials
  • Assessments and lab activities
  • Software infrastructure supporting the course

Working with Profs. Kevin Leyton-Brown, Giulia Toti, Elham Khoda, and the AI 100 faculty team.

Graduate Teaching Assistant

UBC · January 2024–present

  • AI 422 · Intelligent Systems
  • AI 322 · Foundations of Artificial Intelligence
  • CPEN 311 · Digital Systems Design