About Me

Hi! My name is Ahmed Attia and I’m on an endeavor to satisfy my curiosity and contribute to the betterment of humanity through research in Artificial Intelligent systems. I recieved my MSc from Mohamed bin Zayed University of Artificial Intelligence: MBZUAI under the supervision of Prof. Alham Fikri Aji and Prof. Salem Lahlou. I’m mainly interested in Reinforcement Learning, Reliablity, Robustness of ML Systems and AI for Science.

STEP 000

ACTION explore

REWARD +0.42

SUPREME MISSION maximize humanity's long-term utility

Publications

1. Improving Low-Resource Machine Translation via Round-Trip Reinforcement Learning
   Ahmed Attia and Alham Fikri Aji. arXiv preprint arXiv:2601.12535, 2026.

2. Accurate and Diverse LLM Mathematical Reasoning via Automated PRM-Guided GFlowNets
   Adam Younsi, Ahmed Attia, Abdalgader Abubaker, Mohamed El Amine Seddik, Hakim Hacid, and Salem Lahlou. arXiv preprint arXiv:2504.19981, 2025.

View all publications on Google Scholar.

Blog

Exploring Ordered Sampling in Generative Flow Networks

December 3, 2025

Can choosing which trajectory lengths a GFlowNet sees first improve exploration? We study two simple curricula and find that learning from short trajectories before long ones is especially effective when rewards are sparse.

---

Page design by Ankit Sultana and edited by Ahmed Attia