About

I am a PhD candidate in Computer Science at UMass Amherst, advised by Shlomo Zilberstein. I study the alignment of agentic systems. My recent work uses large language models to infer intent from unstructured instructions in both single and multiagent systems.

News

  • Nov 2025 Full paper accepted at AAAI 2026.
  • Dec 2024 Full paper accepted at AAAI 2025.
  • Dec 2023 Passed PhD portfolio with distinction.

Patents

  1. Dynamic Refinement of Custom Classes Using Zero-Shot Images. Under review.
  2. Vehicle Decision Making Using Sequential Information Probing. US Patent App. 18/429,196.

Publications

Journal

  1. Samer B. Nashed, Saaduddin Mahmud, Claudia V. Goldman, and Shlomo Zilberstein. Causal Explanations for Sequential Decision Making Under Uncertainty: Foundations and Analysis. Journal of Artificial Intelligence Research, 2025. Also accepted to AAAI 2026 (Journal Track). [JAIR] [PDF]

Conferences

  1. Saaduddin Mahmud, Mason Nakamura, Kyle Hollins Wray, and Shlomo Zilberstein. Inference-Aware Prompt Optimization for Aligning Black-Box Large Language Models. AAAI 2026. [PDF]
  2. Saaduddin Mahmud, Mason Nakamura, and Shlomo Zilberstein. MAPLE: A Framework for Active Preference Learning Guided by Large Language Models. AAAI 2025. [PDF]
  3. Saaduddin Mahmud, Marcell VazquezChanlatte, Stefan Witwicki, and Shlomo Zilberstein. Explaining the Behavior of POMDP-based Agents Through the Impact of Counterfactual Information. AAMAS 2024. [PDF]
  4. Connor Basich*, Saaduddin Mahmud*, and Shlomo Zilberstein. Learning Constraints on Autonomous Behavior from Proactive Feedback. IROS 2023. [PDF]
  5. Saaduddin Mahmud, Sandhya Saisubramanian, and Shlomo Zilberstein. Explanation-Guided Reward Alignment. IJCAI 2023. [PDF]
  6. Saaduddin Mahmud, Connor Basich, and Shlomo Zilberstein. Semi-Autonomous Systems with Contextual Competence Awareness. AAMAS 2023. [PDF] [ACM]
  7. Saaduddin Mahmud, Md. Mosaddek Khan, Moumita Choudhury, Long Tran-Thanh, and Nicholas R. Jennings. Learning Optimal Temperature Region for Solving Mixed Integer Functional DCOPs. IJCAI 2021. [PDF]
  8. Saaduddin Mahmud, Moumita Choudhury, Md. Mosaddek Khan, Long Tran-Thanh, and Nicholas R. Jennings. AED: An Anytime Evolutionary DCOP Algorithm. AAMAS 2020. [PDF]
  9. Moumita Choudhury, Saaduddin Mahmud, and Md. Mosaddek Khan. A Particle Swarm Based Algorithm for Functional Distributed Constraint Optimization Problems. AAAI 2020. [PDF]

Workshops, Abstracts & Posters

  1. Saaduddin Mahmud, Eugene Bagdasarian, and Shlomo Zilberstein. CoLLAB: A Framework for Designing Scalable Benchmarks for Agentic LLMs. Scaling Environments for Agents Workshop at NeurIPS 2025. [OpenReview] [GitHub] [Project]
  2. Saaduddin Mahmud, Sandhya Saisubramanian, and Shlomo Zilberstein. Verification and Validation of AI Systems Using Explanations. AAAI Fall Symposium 2024. [PDF]
  3. Saaduddin Mahmud, Sandhya Saisubramanian, and Shlomo Zilberstein. REVEALE: A Framework for Reward Verification and Learning. SafeAI Workshop at AAAI 2023. [PDF]
  4. Saaduddin Mahmud, Samer B. Nashed, Claudia V. Goldman, and Shlomo Zilberstein. Estimating Causal Responsibility for Explaining Autonomous Behavior. EXTRAAMAS at AAMAS 2023. [Springer] [PDF]
  5. Rafid Ameer Mahmud, Fahim Faisal, Saaduddin Mahmud, and Md. Mosaddek Khan. A Simulation Based Online Planning Algorithm for Multi-Agent Cooperative Environments. AAMAS 2022 (Extended Abstract). [PDF]

Preprints & Under Review

  1. Mason Nakamura, Abhinav Kumar, Saaduddin Mahmud, Sahar Abdelnabi, Shlomo Zilberstein, and Eugene Bagdasarian. Terrarium: Revisiting the Blackboard for Studying Multi-agent Attacks. Under review, ICLR 2026. [PDF] [GitHub] [Project]
  2. Saaduddin Mahmud, Dorian Benhamou-Goldfajn, and Shlomo Zilberstein. Distributed Multi-agent Coordination Using Multimodal Foundation Models. arXiv 2025. [arXiv] [PDF]