Mahi (Mohammad Mahdi) Rahimi

Postdoctoral Fellow, InnoCORE AITA @ DGIST (from Oct 2025) • Ph.D. (KAIST, Aug 2025)
Emails
mahi@dgist.ac.kr
Office
DGIST, 333 Techno jungang-daero, Daegu, South Korea

Bio

I am a postdoctoral fellow (InnoCORE AITA) at DGIST supervised by Prof. Daewon Seo (from Oct 2025), and completed my Ph.D. in Electrical Engineering at KAIST with Prof. Jaekyun Moon (Aug 2025). My research centers on communication-efficient federated learning and low-rank adaptation for foundation models, with broader interests across MARL, geometric DL, and evolutionary computation. I co-authored EvoFed (NeurIPS 2023) and am developing model-agnostic projection methods for message-efficient FL at scale.

Before this, I completed my undergraduate studies at the Tehran Polytechnic (Amirkabir University of Technology), where I was advised by Prof. MohammadAzam Khosravi and Prof. MohammadMehdi Ebadzadeh on RoboCup SSL platform and multi-agent learning. Between 2019 and 2020, I worked at BRTel Co. on Business Intelligence and BigData solutions for MCI Telecommunication.

News

  • Oct 2025 — Started as Postdoctoral Fellow @ DGIST InnoCORE AITA (ILT Lab, Prof. Dae-Won Seo).
  • Aug 2025 — Ph.D. completed at KAIST (MoonLab).
  • Jun 2025Best Project (Naver-Intel-KAIST) for Gaudi-based message-efficient FL.
  • Dec 2023EvoFed appeared at NeurIPS 2023.
  • Sep 2020 — Joined MoonLab @ KAIST.
  • Sep 2020 — Completed B.Sc. in Software Engineering, AUT.
  • Jul 2018 — Paper accepted at the Journal of Open Source Software (OPEM).
  • Feb 2015 — Joined AUT as Talented Student.

Publications

EvoFed: Leveraging Evolutionary Strategies for Communication-Efficient Federated Learning
Mohammad Mahdi Rahimi, Hasnain Bhatti Irshad, Younghyun Park, Humaira Kousar, Jaekyun Moon — NeurIPS 2023 • Code
OPEM: Open-Source PEM Cell Simulation Tool
Haghighi, Askari, Hamidi, Rahimi — JOSS 2018 • Code
RoboCup SSL Extended Team Description Papers

Academic Service

  • ICLR 2026 — Reviewer
  • ICML 2025 — Reviewer
  • NeurIPS (2022-2025) — Reviewer
  • AI WorldCup (2019) — Advisory Board
  • FIRA RoboWorld Cup (2017–2019) — League Chair

Honors & Awards

  • InnoCORE Postdoctoral Fellowship (₩90M/yr)
  • Naver-Intel-KAIST Best Project, 2025
  • EurAI Full-Travel Grant, ACAI, 2019
  • Alibaba Tianchi (Top 14/3224), 2018
  • Russian AI Challenge (Top 40/780), 2018
  • RoboCup / FIRA podiums & awards (2014–2018)

Education

  • KAIST — Ph.D., Electrical Engineering (2020–2025). Advisor: Jaekyun Moon. Thesis: Communication-Efficient Federated Learning.
  • Amirkabir University of Technology — B.Sc., Computer Engineering (2015–2020). Advisor: M.M. Ebadzadeh. Thesis: MARL in Soccer Robots.

Contact

Email: mahi@dgist.ac.kr
GitHub: mahi97 • Scholar: profile
LinkedIn: mahi97