Wrapped up nearly four years as a Research Engineer at QCRI.
"Said Aloud, Read Different" accepted at Interspeech 2026.
"Once Correct, Still Wrong", on counterfactual hallucination in multilingual VLMs, accepted to Findings of ACL 2026. The M2CQA benchmark is on Hugging Face.
Fanar 2.0 technical report released.
Survey on multilingual knowledge editing accepted at EMNLP 2025.
"An Exploration of Knowledge Editing for Arabic" accepted at ArabicNLP 2025.
"From Words to Waves" published at Interspeech 2025.
AraDiCE, benchmarks for Arabic dialects and culture in LLMs, published at COLING 2025.
Fanar, an Arabic-centric generative AI platform, released.
"Exploring Alignment in Shared Cross-lingual Spaces" published at ACL 2024.
LAraBench and LLMeBench published at EACL 2024.
"Can LLMs Facilitate Interpretation of Pre-trained Language Models?" published at EMNLP 2023.