Azmine Toushik Wasi
黑料社
I am an ML/NLP researcher focused on building reliable, safe, and human centered AI systems, with healthcare as my primary application domain. My research explores how foundation models reason and why they fail, how machine learning can advance our understanding of biological systems and clinical outcomes, and how people understand, trust, and use AI in practice.
I approach ML/AI from three complementary perspectives:
- the model and systems side, developing and evaluating foundation models and agentic systems for evidence grounded reasoning, structured retrieval, claim verification, factual reliability, uncertainty estimation, adaptive decision making, alignment, and mechanistic interpretability;
- the biomedical-clinical side, developing machine learning methods using molecular, biological, and clinical data to study diseases, treatments, and patient outcomes; and
- the human side, investigating how people interact with AI systems and how to improve their safety, transparency, reliability, and fairness across languages, cultures, and healthcare settings.
I collaborated with researchers at MBZUAI (UAE), QCRI (Qatar), Microsoft Research (USA), MIT Media Lab (USA), SRI (Canada), KSU (Saudi Arabia), Pi School (Italy), Mila (Canada), Cohere Labs (Canada), CIOL (Bangladesh), CMU (USA), and Hanyang University (South Korea). My research has appeared in leading venues including ICLR, ICML, NeurIPS, ACL, WWW, EMNLP, and ACM FAccT, alongside associated workshops. Coming from a non CS undergraduate background, I published first or last author papers at ICLR, ICML, and NeurIPS in 2026, completing the three major ML conferences before beginning graduate study. I regularly review for leading AI and ML conferences and journals, including NeurIPS, ICLR, ICML, UAI, AISTATS, TMLR, AAAI, and ACL ARR, and received ICML Gold Reviewer recognition in 2026. I am also a Kaggle Grandmaster, with an Erd艖s number and Dijkstra number of 4.
University roles and responsibilities
- PGR Student, People-Centred AI
My qualifications
Overall CGPA: 3.74/4.00; Final year: 3.94/4.00
ResearchResearch interests
I'm a ML researcher focused on reliable, safe, and human-centered AI, with healthcare as a primary application domain. My research investigates foundation model reliability, evidence-grounded reasoning, safety, alignment, and human trust, aiming to develop reliable AI systems that support informed decision-making under uncertainty. My research interests are:
- Foundation Models, Reasoning, Alignment, and Agents: Evidence-grounded reasoning across domains, retrieval-augmented generation, factual verification, uncertainty estimation, hallucination detection, reasoning faithfulness, model calibration, agent design and evaluation, tool use, alignment, and mechanistic interpretability of model behavior and failures.
- Computational Biology and Health Informatics: Molecular and clinical modeling, biomedical knowledge integration, disease and treatment prediction, clinical evidence grounding, bio-safety, and biomedical risk assessment.
- Human-AI Interaction, Safety, and Fairness: Human oversight, trust, transparency, AI governance and policy, and fairness across languages, cultures, and healthcare settings.
Research projects
Toward Reliable and Transparent Foundation Models for Domain-Specific ApplicationsCurrent foundation models in large language models still struggle with factual reliability, especially when they are used in specialized or high-stakes domains. Although models such as GPT, LLaMA, Claude, and Gemini can generate fluent and convincing responses, they may still produce hallucinated or unsupported information. Recent research shows that hallucination remains a central limitation even in retrieval-augmented systems, where models are connected to external sources, because errors can occur during retrieval, reasoning, or generation. This creates a research gap in developing foundation models that can not only generate accurate language, but also verify their own outputs, express uncertainty, and provide traceable evidence for claims. Therefore, this proposal focuses on improving the factual reliability and transparency of foundation models in large language models, particularly for domain-specific applications where incorrect outputs can lead to serious consequences.
Research interests
I'm a ML researcher focused on reliable, safe, and human-centered AI, with healthcare as a primary application domain. My research investigates foundation model reliability, evidence-grounded reasoning, safety, alignment, and human trust, aiming to develop reliable AI systems that support informed decision-making under uncertainty. My research interests are:
- Foundation Models, Reasoning, Alignment, and Agents: Evidence-grounded reasoning across domains, retrieval-augmented generation, factual verification, uncertainty estimation, hallucination detection, reasoning faithfulness, model calibration, agent design and evaluation, tool use, alignment, and mechanistic interpretability of model behavior and failures.
- Computational Biology and Health Informatics: Molecular and clinical modeling, biomedical knowledge integration, disease and treatment prediction, clinical evidence grounding, bio-safety, and biomedical risk assessment.
- Human-AI Interaction, Safety, and Fairness: Human oversight, trust, transparency, AI governance and policy, and fairness across languages, cultures, and healthcare settings.
Research projects
Current foundation models in large language models still struggle with factual reliability, especially when they are used in specialized or high-stakes domains. Although models such as GPT, LLaMA, Claude, and Gemini can generate fluent and convincing responses, they may still produce hallucinated or unsupported information. Recent research shows that hallucination remains a central limitation even in retrieval-augmented systems, where models are connected to external sources, because errors can occur during retrieval, reasoning, or generation. This creates a research gap in developing foundation models that can not only generate accurate language, but also verify their own outputs, express uncertainty, and provide traceable evidence for claims. Therefore, this proposal focuses on improving the factual reliability and transparency of foundation models in large language models, particularly for domain-specific applications where incorrect outputs can lead to serious consequences.
Teaching
- Global Teaching Assistant (Volunteer), How to Grow Almost Anything (HTGAA), MIT Media Lab (Aug 2025鈥揊eb 2026)
- Supported a computational genetics, bioengineering, and biodesign course for hundreds of participants worldwide; advised individual and group projects. Contributed to lab automation projects involving Ginkgo and Opentrons.
- Lead Organizer and Instructor, CIOL Winter ML Bootcamp (Dec 2024鈥揊eb 2025)
- Designed and taught 70+ attendees on data analysis, ML, NLP, vision, AI agents, and research methods.
Sustainable development goals
My research interests are related to the following:
Publications
馃摪 Conference Papers
Azmine Toushik Wasi, Wahid Faisal, Abdur Rahman, Mahfuz Anik, Munem Shahriar, Mohsin Topu, Sadia Tasnim Meem, Rahatun Nesa Priti, Sabrina Afroz Mitu, Md. Iqramul Hoque, Shahriyar Zaman Ridoy, Mohammed Eunus Ali, Majd Hawasly, Mohammad Raza, Md Rizwan Parvez
ICLR 2026
Azmine Toushik Wasi, Mst Rafia Islam, Mahfuz Ahmed Anik, Md Manjurul Ahsan, TH Rafi, Dong-Kyu Chae
ICML 2026: Position Papers (Spotlight, Top 5%) SPOTLIGHT
Azmine Toushik Wasi*, Shahriyar Zaman Ridoy*, Koushik Tonmoy, Kinga Tshering, S. M. M. Hasan, Wahid Faisal, T. Mohiuddin, Md Rizwan Parvez
ICML 2026
Ishtiak Mahmud Saad, Mominul Islam, Shahriyar Zaman Ridoy, Md Manjurul Ahsan, Azmine Toushik Wasi$
NeurIPS 2026: Main Technical Track
Shahriyar Zaman Ridoy, S. M. Muhtasimul Hasan, Azmine Toushik Wasi, Kinga Tshering, Koushik Ahamed Tonmoy, Md Mosaddek Khan
EMNLP 2026 (Findings, Top 30%)
Md Rayhanul Masud, Azmine Toushik Wasi, Salman Rahman, Md Rizwan Parvez
EMNLP 2026 (Findings, Top 30%)
Azmine Toushik Wasi
| (Top 20%, Oral) ORAL
Azmine Toushik Wasi, Mohsin Mahmud Topu, Mahfuz Ahmed Anik, Md Manjurul Ahsan
(Top 22.5%, Oral) ORAL
Md Asif Bin Syed, Md Younus Ahamed, Azmine Toushik Wasi$
ICML 2026: Position Papers
Azmine Toushik Wasi, Wahid Faisal, Mst Rafia Islam, Md Rizwan Parvez
ACL 2026 Findings
Pedro Ortiz Suarez, Laurie Burchell, ..., Azmine Toushik Wasi, ..., Kenton Murray, Sarah K. K. Luger
ACL 2026 Main
Azmine Toushik Wasi, Mahdiya Rahman Sukanya
Azmine Toushik Wasi, Mahir Absar Khan, Rahatun Nesa Priti, Abdur Rahman, Mst Rafia Islam, Miskat Tahmin Ali
Israfel Salazar, Manuel Fern谩ndez Burda, Shayekh Bin Islam, ..., Azmine Toushik Wasi, ..., Sara Hooker, and Marzieh Fadaee
ICLR 2026
Azmine Toushik Wasi, Mst Rafia Islam, Wahid Faisal, Abdur Rahman
Late Breaking Results (LBR: Posters)
Azmine Toushik Wasi, Wahid Faisal, Mst Rafia Islam
Late Breaking Results (LBR: Posters)- Spatial Reasoning in Multimodal Foundation Models: A Survey of Spatial Understanding in LLMs and VLMs
Azmine Toushik Wasi, Md. Masudur Rahman, Abdur Rahman, Mohsin Mahmud Topu, Mohammed Eunus Ali, Riashat Islam, Md Rizwan Parvez
AACL-IJCNLP 2026: Main - Geo-spatial and Geo-temporal Reasoning in Vision-Language and Large Language Models: A Review
Shahriyar Zaman Ridoy, Azmine Toushik Wasi, Koushik Ahamed Tonmoy, Md Rizwan Parvez
AACL-IJCNLP 2026: Main - Large Language and Vision-Language Models as World Models: A Survey
Azmine Toushik Wasi, Sadia Tasnim Meem, Riashat Islam, Md Rizwan Parvez
AACL-IJCNLP 2026: Findings - Is There a Dr. House in the Machine? A Survey of Clinical and Diagnostic Reasoning in LLM-based Medical Agents
Azmine Toushik Wasi, Abdur Rahman, Riashat Islam, Md Rizwan Parvez
AACL-IJCNLP 2026: Findings - Position: Real-World Clinical NLP Robustness Requires Messy, Multimodal, Longitudinal, Privacy-Preserving Corpora
Azmine Toushik Wasi, Shahriyar Zaman Ridoy
AACL-IJCNLP 2026: Findings
Azmine Toushik Wasi, Omid Reza Heidari*, MD Mohaymen Ul Anam*, Abdur Rahman*, Amit Agarwal, Hitesh Laxmichand Patel, Bhargava Kumar, TH Rafi, Dong-Kyu Chae
PAKDD 2026: DSFA
Mahfuz Ahmed Anik*, Azmine Toushik Wasi*, Mohsin Mahmud Topu, TH Rafi, Dong-Kyu Chae
PAKDD 2026: DSFA
Shahriyar Zaman Ridoy*, Azmine Toushik Wasi*, Koushik Ahamed Tonmoy, TH Rafi, Dong-Kyu Chae
ACM FAccT 2026
Azmine Toushik Wasi, Mahfuz Ahmed Anik, Md Shafikul Islam, Md Manjurul Ahsan
(Health Systems Track > Best Track Paper : Second Place)
Mahfuz Ahmed Anik, Azmine Toushik Wasi, Md Shafikul Islam, Md Manjurul Ahsan
Azmine Toushik Wasi, Raima Islam*, Mst Rafia Islam*, Farig Sadeque, TH Rafi, Dong-Kyu Chae
CSCW 2025 Posters
Azmine Toushik Wasi, Mst Rafia Islam, Abdur Rahman, Tawfia Yeasmin, Amit Agarwal, Hitesh Laxmichand Patel, TH Rafi, Dong-Kyu Chae
CSCW 2025 Posters
Wahid Faisal*, Azmine Toushik Wasi*, Drishti Sharma*, Mahfuz Ahmed Anik, TH Rafi, Dong-Kyu Chae
CSCW 2025 Posters
Azmine Toushik Wasi, Raima Islam, Mst Rafia Islam, Farig Sadeque, TH Rafi, Dong-Kyu Chae
(Short Paper)
Angelika Romanou, Negar Foroutan, Anna Sotnikova,... , Azmine Toushik Wasi, ... , Marzieh Fadaee, Sara Hooker, Antoine Bosselut
ICLR 2025 (Spotlight, Top 5.1%) SPOTLIGHT
Azmine Toushik Wasi, TH Rafi, Dong-Kyu Chae
DASFAA 2025 (Full Paper, Short Oral, Top 10%) SPOTLIGHT
Ferdinand Lenchau Bit, Iman Khaleda binti Zamri, Azmine Toushik Wasi$, TH Rafi, Dong-Kyu Chae
DASFAA 2025 (Short Paper Paper, Poster)
Azmine Toushik Wasi, Mahfuz Ahmed Anik, Abdur Rahman, Md Iqramul Hoque, Wahid Faisal, and A. M. M. Mukaddes
ICCIT 2025
Abdur Rahman, Mahfuz Ahmed Anik, Shahoriar Muttaki Utshaw, Azmine Toushik Wasi$, and A. M. M. Mukaddes
ICCIT 2025
Azmine Toushik Wasi*, TH Rafi*, Raima Islam, Karlo 艩erbetar, Dong-Kyu Chae
ACCV 2024
Azmine Toushik Wasi, TH Rafi, Raima Islam, Dong-Kyu Chae
Azmine Toushik Wasi
Azmine Toushik Wasi, Karlo Serbetar, Raima Islam, TH Rafi, Dong-Kyu Chae
Azmine Toushik Wasi, TH Rafi, Dong-Kyu Chae
馃摪 Journal Papers
Azmine Toushik Wasi, TH Rafi, Raima Islam, Serbetar Karlo, Dong-Kyu Chae
IEEE/ACM Transactions on Computational Biology and Bioinformatics (TCBB) (Q1/Q2, IF: 4.5, CiteScore: 9.7)
Hugo MacDermott-Opeskin, Cas Wognum, ..., Azmine Toushik Wasi, ..., James Fraser, John D. Chodera
Journal of Chemical Information and Modeling (JCIM) (Q1, IF 5.3)
A. S. A.*, Azmine Toushik Wasi*, Mahfuz Ahmed Anik, MD Shafikul Islam, and Mohamed Kamel Hadj-Kali
International Journal of Production Research (IJPR) (Q1)
Azmine Toushik Wasi, Sadia Tasneem Meem
ACM AI Letters
Mohsin Mahmud Topu#, Azmine Toushik Wasi#, Mahfuz Ahmed Anik, MD Manjurul Ahsan
Intelligent Transportation Infrastructure (Q1)
Mahfuz Ahmed Anik, Abdur Rahman, MD Shafikul Islam, Md Isfar Khan, Md Manjurul Ahsan, Azmine Toushik Wasi$#
Digital Twins and Applications
MD Shafikul Islam, Mahathir Mohammad Bappy, Saifur Rahman Tushar, Christian Zamiela, Azmine Toushik Wasi
International Journal of Advanced Manufacturing Technology (Q1)
Syeda Kumrun Nahar, Mohammad Muhshin Aziz Khan, Pritidipto Paul Chowdhury, Azmine Toushik Wasi, M. Morad Ali & M. Rifat Rahman
International Journal of Advanced Manufacturing Technology (Q1)
MD Shafikul Islam, Mahfuz Ahmed Anik, Azmine Toushik Wasi, Mahathir Mohammad Bappy
IISE Transactions on Healthcare Systems Engineering (Q2)
馃搩 Pre-prints / Works in Review
Mahir Shahriar Tamim*, Md. Samiul Alim*, Azmine Toushik Wasi, Shahriyar Zaman Ridoy, Meharun Nesa, M. Abu Yousuf, Alex Lamb, M. Ali Moni
Azmine Toushik Wasi, Mahir Absar Khan, Abdur Rahman, Wahid Faisal, Sukanta Saha, Saimon Bhuiyan, Mahdiya Rahman Sukanya, Rahatun Nesa Priti, Md. Iqramul Hoque, Munem Shahriar, Raima Islam, Sanatan Sushil, Shahriyar Zaman Ridoy, Kazi Rajwan Sultan, Md Tanzib Hosain, MD Shafikul Islam, Dong-Kyu Chae, Md Manjurul Ahsan, Md Rizwan Parvez
Azmine Toushik Wasi, Adinath Madhavrao Dukre, Md Manjurul Ahsan, Muhammad Awais, Sara Atito
Srikar Vardhan, Adinath Madhavrao Dukre, Azmine Toushik Wasi, Imran Razzak
Adinath Madhavrao Dukre, Govinda Kolli, Yifan Lu, Ziyun Zou, Azmine Toushik Wasi, Sara Atito, Behzad Bozorgtabar, Dwarikanath Mahapatra, Muhammad Awais, Imran Razzak
Wahid Faisal*, Azmine Toushik Wasi*, Mohammed Eunus Ali, Md Rizwan Parvez
Azmine Toushik Wasi*, Wahid Faisal, Mohammed Eunus Ali*, Md Rizwan Parvez
Azmine Toushik Wasi, Ankan Deria, Adinath Madhavrao Dukre, Sara Atito, Muhammad Awais, Imran Razzak
Azmine Toushik Wasi, Md Tanzib Hosain, Adinath Madhavrao Dukre, Md Manjurul Ahsan, Muhammad Awais, Sara Atito, Imran Razzak
Vil茅m Zouhar, Niyati Bafna, Mukund Choudhary, Maike Z眉fle, ... , Azmine Toushik Wasi, ... (and 255 others)
Azmine Toushik Wasi, Md Manjurul Ahsan
Adinath Madhavrao Dukre, Yifan Lu, Govinda Kolli, Ziyun Zou, Youssef Mohamed, Azmine Toushik Wasi, Yutong Xie, Shoaib Jameel, Imran Razzak
Noor Mairukh Khan Arnob*, Azmine Toushik Wasi*, Md Noushad Jahan Ramim, Wahid Faisal, Labiba Akram Pritha, Saiyara Mahmud, Md Rizwan Parvez
Azmine Toushik Wasi, MD Shafikul Islam Sohan, Adipto Raihan Akib, Mahathik Muhammad Bappy
馃捈 Academic Services
- Conference or Journal Reviewer
- AI/ML: ICML (), ICLR, NeurIPS, UAI, AISTATS, AAAI.
- NLP: ACL ARR 2024-Present (ACL, EMNLP, NAACL, EACL, COLING), []; COLING, and Industry Tracks.
- CV: WACV.
- Bio/Medical/Health Infomatics: MICCAI, MLCB, IEEE EMBS BHI.
- HCI/HAI: CHI, CSCW, UbiComp, TEI, ICM, IDC, ACM EAAMO.
- Journal Reviewer
- Transactions on Pattern Analysis and Machine Intelligence (T-PAMI) (IF 18.6)
- Transactions on Machine Learning Research (TMLR)
- IEEE Transactions on Affective Computing (TAFFC) (IF 9.8)
- Complex & Intelligent Systems (Springer Nature) (IF 5.6)
- ACM Transactions on Autonomous and Adaptive Systems (TAAS) (IF 3.8)
- Organizer
- Four workshops at NeurIPS 2024, ICML 2024, and NeurIPS 2025, ICML 2026 and six socials across NeurIPS, ICML, ICLR, AISTATS, and ACL, as a Affinity Workshop Steering Committee Member and Core Organizer.