Hi, I’m Mohammad. I grew up in Qom, Iran, and I’m now in Toronto doing my MASc at York University with Prof. Jack Jiang.
I work on making LLM-powered software something you can trust: reproducible runs, honest benchmarks, and coding agents that don’t burn money on dead ends. That’s split between my research at York and an internship at Grimlock, where I test ways to get agents like Claude Code to solve more tasks for less.
Current Status
Oct. 2026- Waiting on reviews for our FSE 2027 submission.
- Early work on a follow-up about how reliable software engineering benchmarks really are.
- Running weekly coding-agent experiments at Grimlock.
- Learning French, slowly.
Recent News
All news »- [2026/10] Submitted our paper "Taming Non-Reproducibility: A Multi-Layered Framework for Divergence Investigation and Reproducible Execution in LLM-Powered Systems" to FSE 2027. It is now under review.
- [2026/03] Still working on our paper "Taming Non-Reproducibility: A Multi-Layered Framework for Divergence Investigation and Reproducible Execution in LLM-Powered Systems", with some interesting results coming out of it along the way.
- [2026/03] Updated my personal website and built a personal profile chatbot.
Publications
- S. Mohammad Hejazi and Z. M. (Jack) Jiang. Taming Non-Reproducibility: A Multi-Layered Framework for Divergence Investigation and Reproducible Execution in LLM-Powered Systems. Under review, 2026.
- B.Sc. Thesis: Building a Recommendation System Including User-Specific Recommendation, Similar Items, and Cold Start. Amirkabir University of Technology, 2023. GitHub · PyPI
Experience
Full details »- Research Intern, Grimlock Consulting Sep. 2026 - Present
- Research Assistant, York University Jan. 2025 - Present
- Teaching Assistant, York University Jan. 2025 - Apr. 2026
- Teaching Assistant, Amirkabir University of Technology Feb. 2022 - Jun. 2022
- Intern, Qom Province Electricity Distribution Company Jun. 2021 - Sep. 2021
Selected Projects
All projects »KabirRec
Bachelor Thesis · Kubernetes · Docker · Python
A recommendation system on Kubernetes plus a Python library for building recommenders from scratch, later used in a larger AI-as-a-service framework.
URL Shortener
Kubernetes · Docker · MongoDB · Python
An API server on Kubernetes that shortens URLs and stores them in MongoDB, load balanced and scalable on demand.
AI Agents in Pac-Man
Search · Adversarial Games · Reinforcement Learning · Python
Agents for Pac-Man using BFS, DFS, UCS, and A*, then Minimax and Expectiminimax, then Value Iteration and Q-Learning.
Education
Full details »York University
Jan. 2025 – Present
MASc in Electrical & Computer Engineering
Toronto, Canada
Amirkabir University of Technology
Sept. 2018 – Jul. 2023
BSc in Computer Engineering
Tehran, Iran
Beyond Work
More about me »- ♟️ Used to play a lot of chess, with a peak Chess.com rating of 1740.
- 📷 In my free time, I make and edit videos on YouTube.
- 🏋️♂️ Working out regularly. Best: 61 push-ups in 2 minutes, and 25 pull-ups in one set.