I'm Mohammad Reza

As an M.Sc. student at Mila - Québec AI Institute and Université de Montréal, I currently hold a visiting researcher position at ServiceNow. My past experience includes an internship at EPFL and AI research roles at companies. I've earned my B.Sc. in Computer Science from Sharif University of Technology.
My scientific interests center on understanding how intelligence is grounded in computation, specifically developing agents capable of reasoning in complex situations. I'm also dedicated to bringing AI solutions to real-world decision-making problems and collaborate with startups seeking to integrate AI into their systems.
I am not committed to or against the development of artificial general intelligence for its own sake, but rather envision the creation of machines that can assist us in addressing and solving our problems. Furthermore, I strongly advocate for an active exploration of the ways in which AI shifts power dynamics and fundamentally transforms our world.

My CV

Publications

Mastering Memory Tasks with World Models

Mohammad Reza Samsami*, Artem Zholus*, Janarthanan Rajendran, Sarath Chandar
ICLR 2024 (oral, top-1.2%)

Mining the Diamond Miner: Mechanistic Interpretability on the Video PreTraining Agent

Sonia Joseph*, Artem Zholus*, Mohammad Reza Samsami*, Blake A. Richards
ATTRIB Workshop, NeurIPS 2023

Causal Imitative Model for Autonomous Driving

Mohammad Reza Samsami, Mohammadhossein Bahari, Saber Salehkaleybar, Alexandre Alahi
Under Review Preprint

Recognizing Arrow of Time in the Short Stories

Fahimeh Hosseini, Hossein Fooladi, Mohammad Reza Samsami
WiNLP Workshop, ACL 2019

Distributed Deep Reinforcement Learning: An Overview

Mohammad Reza Samsami, Hossein Alimadad
Preprint

Experiences

Education

  • Mila - UdeM
    2021 - Pres.

    I am currently pursuing my Master of Science at Mila - Québec AI Institute and Université de Montréal working with Profs. Sarath Chandar and Irina Rish. My work is centered on world models in reinforcement learning, with additional investigations into causality and mechanistic interpretability within the same domain.

  • Sharif University of Technology
    2016 - 2021

    I completed my undergraduate studies in Computer Science at the Sharif University of Technology, where I ranked top 0.3% in Iran’s national entrance exam. I was Research Assistant under Prof. Salehkaleybar, and my research contributions were focused on advancing the domains of reinforcement learning, imitation learning, and causal learning.

  • National Olympiad in Informatics
    2013 - 2015

    My journey into the field of Computer Science began with rigorous preparation for the Olympiad in Informatics during high school. This intensive training honed my abilities to tackle complex combinatorial questions, algorithmic problems, and competitive programming. Consequently, I earned a silver medal in the 25th Iran’s Nationwide Olympiad in Informatics.

Research & Work

  • ServiceNow
    2023 - Pres.

    As a Visiting Researcher at ServiceNow Research in Montreal, my primary focus lies in improving the sample efficiency of reinforcement learning from human feedback (RLHF) for Large Language Models (LLMs), by incorporating principles of causality in reward modeling.

  • EPFL
    2020 - 2021

    During a research internship at EPFL, I worked under the guidance of Profs. Alahi and Salehkaleybar. My research revolved around leveraging causality to enhance the reliability of imitation learning methods, specifically for the betterment of autonomous driving systems.

  • MetoData Company
    2021

    As a data scientist, I contributed to our team's mission of building autonomous agents for quantitative finance by creating a crypto trader using deep reinforcement learning, specializing in Bitcoin-to-USD trading. The resulting product, caters to small and medium-sized enterprises.

  • Governance Notebook
    2021

    I co-founded Governance Notebook, a think tank dedicated to offering well-rounded perspectives to enhance efficiency and transparency in policy-making processes. Additionally, I have served as the head of its executive team, steering our collective efforts towards impactful contributions.

  • Shenakht Pajouh
    2017 - 2020

    At Shenakht Pajouh, a startup specializing in AI and cognitive science, I held the positions of researcher and program manager. I was actively involved in a research initiative aimed at acquiring fundamental concepts of linguistics, with the goal of developing AI models capable of language comprehension.

  • Teaching
    2016 - 2021

    Following high school, I spent three years teaching algorithm design, programming, and graph theory to students as an Olympiad in Informatics instructor. Subsequently, I furthered this educational path by taking on a role as a teaching assistant at the Sharif University.

Extracurricular Activities

  • Shenasa
    2017 - 2019

    I co-founded Shenasa, the Sharif Cognitive Science interdisciplinary community, bridging the fields of CS, Mathematics, EE, Physics, Chemistry, and Philosophy of Science. Furthermore, I was elected to serve on its inaugural central council.

  • Hamband Scientific Association
    2017 - 2018

    I served as an elected member of the central council of Hamband, the scientific association of the Computer Science and Mathematics department at Sharif. We organized extracurricular events, including the Data Science Seminar.

  • Data Science Seminar
    2017

    I served as a co-organizer for the first scientific seminar on Data Science at Sharif University, a two-day event that brought together researchers, engineers, and industry leaders for knowledge exchange and networking.