About Me

I am a student research assistant in the Responsible AI and Machine Ethics (RAIME) group, part of the Neuro-Mechanistic Modeling department at DFKI, and an MSc student in Data Science and Artificial Intelligence at Saarland University. My research focuses on the safety of embodied agents. In particular, I work on neuro-symbolic methods that pair LLM-based agents with explicit reasoning about what an agent is permitted to do, so that unsafe actions can be caught before they are executed. Currently I am working on integrating the GRACE framework with embodied-agent safety benchmarks.

Before joining DFKI, I worked as a research assistant at the Max Planck Institute for Informatics on diffusion models for molecules and on LLM-as-a-judge evaluation, and in industry at IBM and Devsinc.

Interests

I am interested in making embodied AI agents behave safely. My current work combines neural components (LLM-based agents) with explicit, symbolic reasoning so that an agent’s decisions can be checked against formal requirements rather than trusted blindly. I am also interested in how such agents should be evaluated, including benchmarks for safety and LLM-based evaluation methods.
  • Embodied Agent Safety
  • Neuro-Symbolic Methods
  • LLM Agents and Evaluation

Education

I completed my bachelor’s degree in computer science at GIKI in Pakistan, where I graduated with distinction. Before coming to Saarbrücken I worked in industry as a data scientist at IBM and as a software engineer at Devsinc. I am now pursuing my MSc in Data Science and Artificial Intelligence at Saarland University.

  • MSc in Data Science and Artificial Intelligence (AI Safety), ongoing

    Supervised by Dr. Kevin Baum at Saarland University

  • BSc in Computer Science, 2022

    Ghulam Ishaq Khan Institute of Engineering Sciences and Technology (GIKI)

    Grade: 3.55/4.00, with Distinction