Md Tanzib Hosain

I am a visiting researcher at MPI-SP, where I am working on socially inclusive and personalized behavioral patterns of LLMs.

My research centers on (1) Foundation Models & Agentic AI: Multi-agent and human-agent knowledge systems, and scientific discovery (2) Reinforcement Learning (RL): Inverse RL, imitation learning, and human-in-the-loop RL, and (3) AI for Science: Learning algorithms for physical systems, human-AI interaction, optimization for scientific discovery, LLMs as search agents, LLMs for math and code generation, AI in medicine, and AI for social science.

I work with Prof. Mohammad Ali Moni (UQ) and Prof. Imran Razzak (MBZUAI) on artificial intelligence in medicine; Prof. Usman Naseem (MQ), Md Mofijul Islam, PhD (Amazon Research), and Tasnim Mohiuddin, PhD (QCRI) on machine translation and linguistic fairness or bias; and Md Rizwan Parvez, PhD (QCRI) and Prof. Mohammed Eunus Ali (Monash) on MAS reasoning.

My research has been published in venues including EMNLP, IJCNLP-AACL, and MIDL, as well as co located workshops in venue including ICML. I regularly review for major AI and ML venues, including ICLR, AAAI, EMNLP, EACL, IJCNLP-AACL, AAMAS, and ACM FAccT.

Email  /  GitHub  /  Scholar  /  LinkedIn

profile photo

Research

project image

When Do Subgroup Specific Acceptance Policies Help in Medical Evidence Verification?


Azmine Toushik Wasi, Md Tanzib Hosain, Adinath Madhavrao Dukre, Md Manjurul Ahsan, Muhammad Awais, Sara Atito, Imran Razzak
AISTATS 2027 In submission

project image

InvariantBench: Can Large Language Models Exhibit Inherent Reasoning Consistently Across Equivalent Transformations?


Azmine Toushik Wasi, Mahir Absar Khan, ..., Md Tanzib Hosain, Md Shafikul Islam, Dong-Kyu Chae, Md Manjurul Ahsan, Md Rizwan Parvez
ICLR 2027 In submission

project image

BRACE: An R Benchmark for Rigorous Biomedical Data Science Code Search


Md Tanzib Hosain, Azmine Toushik Wasi, Salman Rahman, Md Mofijul Islam, Mohammad Ali Moni, Mohammed Eunus Ali, Usman Naseem, Md Rizwan Parvez
ICLR 2027 In submission

project image

HI-REASO: A Multi-Level Multi-Faceted Hindi Mathematical Evaluation Suite for Foundation Models


Md Tanzib Hosain, Rajan Das Gupta, Zannatul Ferdous Tripty, Md Kishor Morol
Findings of EMNLP 2026
code / paper / HF /

project image

Xolver: Generalist Reasoning and Problem Solving through Federated Multi-Agent Dynamics and Holistic Experience Learning


Md Tanzib Hosain, Salman Rahman, Md Kishor Morol, Md Rizwan Parvez
ICML 2026 AI for Math workshop
code / paper / website /

project image

BRAINS: A Retrieval-Augmented Agent for Alzheimer's Detection and Monitoring


Md Kishor Morol, Md Tanzib Hosain, Nafiz Fahad, Md Jakir Hossen, Mohammad Ali Moni
MIDL 2026 Short Papers
paper /

project image

Intrinsic Linguistic Bias in Formal vs. Informal Bengali Pragmatics with Progressive Context Inflation


Md Tanzib Hosain, Md Kishor Morol
Findings of IJCNLP-AACL 2025 (D&I Award)
code / paper /

project image

B-REASO: A Multi-Level Multi-Faceted Bengali Evaluation Suite for Foundation Models


Md Tanzib Hosain, Md Kishor Morol
Findings of EMNLP 2025
code / paper / HF /

Miscellanea

Talks

MPI Spark(l)ing Science, 2026

Academic Service

Program Committee, AAAI 2027
Reviewer, ICLR 2027
Reviewer, ACM FAccT 2027
Reviewer, AAMAS 2027
Reviewer, EACL 2027
Reviewer, EMNLP 2026
Reviewer, IJCNLP-AACL 2026

Award

IJCNLP-AACL 2025 D&I Award

This website's source code uses design from Jon Barron and Leonid Keselman's websites.