Abdullah Hashmat

I am a Computer Science graduate from LUMS specializing in model compression, NLP for low-resource languages, and LLM bias & jailbreak evaluations. Currently working as a Research Assistant for DFKI Germany in collaboration with RPTU in the domain of bioinformatics, exploring the use of LLMs and ML models for molecular property prediction.

My research interests include model compression, AI fairness & Interpretability, reasoning, jailbreaking LLMs, and developing AI solutions for speech processing and analytics.

For my final year thesis, I created PakBBQ, a bias benchmarking QA dataset with 17,180 Urdu/English examples across 8 sociocultural categories. I evaluated 6 multilingual LLMs, revealing cross-linguistic bias gaps and effective bias mitigation strategies. Thus work culminated in a paper accepted at EMNLP 2025 main conference (Core: A*).