About me

I am a Faculty member in the Faculty of Computer Science, Sexton Chair in AI, and Director of HyperMatrix at Dalhousie University, Halifax, Canada. I am an AI researcher with domain expertise in Natural Language Processing and Safe and Trustworthy AI. Moreover, I am a consultant and a mentor blended with entrepreneurial interests.

Research Interest

I work on developing AI systems that are Safe, Reliable, and Trustworthy. My research addresses fundamental questions that enable AI models to achieve these attributes. A few of the research topics that my group is working on are:

• Language generation, interpretability and explainability, generalization and robustness, open weight model safety, model editing and steering, compositionality, conformal prediction and LLM agents

Recent News

Papers

• 2026
  • Areas: Interpretability, Safety, Uncertainity prediction, Low-resource languages
  • Conference: 1 paper (ICML), 3 papers (COLM), 3 papers (ACL)
  • Journal: 1 TMLR
• 2025
  • Conference: 2 papers (ICML), 5 papers (NeurIPs, ICLR, EMNLP, AACL)
  • Workshop: 3 (TAIG @ICML, LockLLM @NeurIPS, WiNLP @EMNLP)
  • Journal: 1 IJMIR
• Accepted Papers 2024: 2 papers (NeurIPs), 2 papers (NAACL), 2 papers (EMNLP)
• Accepted Papers 2023: 3 conference papers (NeurIPS, ICLR, ACL), 1 Journal (JMLR), 3 demo papers (AAAI, ACL, EACL)

Service

• Diversity and Inclusion Chair: NAACL 2027
• Action Editor: TMLR
• Senior AC: ARR 2026, EMNLP 2026
• Area chair: NeurIPS 2026/25/24/23, ARR ²⁰²⁵⁄₂₄, EMNLP 2023
• Reviewer: ICML ²⁰²⁵⁄₂₆, COLM 2026, ArabicNLP 2026, TACL

Talks

• Are LLMs a Good Model of Human-like Understanding? The Challenge of Compositional Learning. Workshop on Prompt Engineering for Pre-Trained Language Models, ACM WebConf’26. Jun. 2026. Keynote
• Latent Concept-Based Explanation of NLP Models. Workshop on Decoding Decisions: Explainability in ML & Sequential Decision Making, CVPR. May 2025
• Navigating Latent Space for Safety, Interpretability and Explainability. New York University Abu Dhabi. May 2025
• Are LLMs a Good Model of Human Thought? The Challenge of Compositional Learning. Atlantic Canada AI Summit. May 2025
• Latent Space Exploration for Safe and Trustworthy AI Models. Representation learning for NLP at ACL. Aug. 2024. Keynote
• Latent Space Exploration for Safe and Trustworthy AI Models. MBZUAI, Abu Dhabi. Aug. 2024
• Latent Space Exploration for Safe and Trustworthy AI Models. AI@Thomson Reuters. Apr. 2024

Misc

• NeuroX Toolkit on pip: https://pypi.org/project/neurox/
• SugarCrepe++ Dataset: https://huggingface.co/datasets/Aman-J/SugarCrepe_pp}