🧑 Career Summary

I am an aspiring AI researcher and a recent graduate of Industrial and Production Engineering from Shahjalal University of Science and Technology.

My research is dedicated to developing human centric, safe, reliable, robust, efficient and capable machine learning systems. Over the past few years, my work has moved toward a single goal: building human-centric AI systems that strengthen healthcare access (bio/medical, clinical, and mental/psychological) and quality. To reach this goal, I am developing myself along three connected directions:
The first is Generative AI, agentic LLM and VLM systems with strong reasoning abilities for both decision support and patient facing communication. The second is computational biology, biomedicine, and data-driven analysis to understand and develop AI systems grounded in scientific foundations. The third is Human-Computer/AI Interaction, ensuring that AI systems remain human-centric, safe, reliable, and unbiased, and that they function effectively across multilingual and multicultural contexts. In my PhD, I aim to work rigorously in one or more of these research directions, and I am actively seeking a PhD or MSc (Research) position beginning in Spring or Fall 2026.

I collaborate with Prof. Chae (HYU) on GenAI, LLM-HCI, Biomolecular ML and Prof. Alshehri (KSU) on Generative AI and health informatics. I also work with Riashat Islam, PhD (Microsoft Research) and Md Rizwan Parvez, PhD (QCRI) on Biomedical AI, GenAI, agents and reasoning; and researchers from Cohere Labs on LLM evaluation, alignment, agents-reasoning and applications, and with Prof. Min Xu (CMU) on biomolecules. Completed HTGAA 2025 (MIT), focusing on protein engineering; and joined as Global TA. Also, I founded CIOL to help young AI researchers, working with Prof. Mukaddes (SUST), Prof. Ahsan (OU) and Prof. Bappy (LSU) on human-centric AI, LLM reasoning and agents, and digital twins for industrial and bio/medical applications.

My research has been published in prestigious venues such as ICLR (A*), WWW (A*), CSCW (A), IEEE BHI (A*), ACCV, DASFAA, and COLING, and related workshops at NeurIPS, ICLR, AAAI, ICML, ACL, EMNLP, NAACL, CHI, IISE, and others. I regularly review for major AI-ML conferences & journals (NeurIPS, ICLR,T-PAMI, ACL, EMNLP, CHI, CSCW and more) and am a Kaggle Grandmaster. I also have 3 years of experience in AI-driven product/content automation and product and project management. I’m a quick learner and highly adaptable, able to lead teams and projects. I enjoy exploring new topics, sharing knowledge, honing skills, experimenting and pushing my limits.

View my Curriculum Vitae →
Feel free to email me (azminetoushik dot wasi at gmail dot com) if you're interested in collaboration or discussing research. I'm open to new ideas and collaborations.

🔍 Research Interests

---

• 🧠 Understanding NLP, Generative AI, Agentic Systems, and Advanced Reasoning
  • Understanding and evaluating how generative models reason, including internal representation learning, planning, and memory for healthcare and beyond.
  • Improving and evaluating model reasoning through structured CoT, uncertainty aware inference, self verification, self correction, and more.
  • Developing agentic AI systems that can plan, retrieve, interact, and adapt, using RAG, PRMs, Long CoT, and human in the loop feedback loops.
  • Advancing trustworthy deployment in high risk environments with interpretability, safety, alignment, and governance of LLMs and LRMs.
  • Exploring AI for Science, knowledge graphs in science & engineering, and supply chain workflows to improve healthcare and HAI outcomes.
• 🧬 Exploring Computational Biology and Biomedical AI Applications
  • Working on structural biology, molecular ML, and computational drug discovery including property prediction, protein discovery, binder design, molecular interactions and healthcare optimization.
  • Applying GNNs, geometric learning, diffusion and flow based models for protein design, molecular graph modeling, and structure function alignment.
  • Exploring AI for Science, Digital Twins, clinical reasoning, and biomedical LLMs to translate discoveries into practical, patient centered decision support.
  • Active interest in Clinical and Mental Health AI to improve healthcare accessibility and treatment quality.
• 🧑‍💻 Interdisciplinary Research on Humans, AI, Language, and Human-Computer/AI Interaction
  • Advancing Human-Centered AI (HCAI) and AI for Good through fair, accessible, and culturally sensitive multilingual and multimodal systems, integrating humans, AI, and language.
  • Designing and evaluating trustworthy AI workflows using structured reasoning, RAG, RL-based refinement, and HCI-driven methods to ensure safe, reliable, and empathetic human-AI interactions.
  • Mitigating generative AI risks such as medical hallucinations, biases, biosafety failures, and harm in clinical, mental health, and decision-support contexts, with attention to governance and policy frameworks.
  • Studying cognitive mechanisms, theory-of-mind-like AI behavior, and human-AI collaboration to improve psychological safety, trust formation, and effective communication in high-stakes and low-resource environments.

📰 News and Updates

---

📄 Selected Publications

---

• SpatiaLab: Can Vision–Language Models Perform Spatial Reasoning in the Wild?
  Azmine Toushik Wasi, Wahid Faisal, Abdur Rahman, Mahfuz Ahmed Anik, Munem Shahriar, Mohsin Mahmud 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'26 In Review

• TimeSpot: Benchmarking Geo-Temporal Understanding in Vision–Language Models in Real-World Settings
  Azmine Toushik Wasi, Shahriyar Ridoy, Koushik Tonmoy, Kinga Tshering, S. M. Hasan, Wahid Faisal, Tasnim Mohiuddin, Md Rizwan Parvez
  ICLR'26 In Review

• Gaussian Regularization in Neural Graph Learning
  Azmine Toushik Wasi, Taki Hasan Rafi, Dong-Kyu Chae
  DASFAA'25 (Full Paper, Short Oral) ▪ GNN

• Dialectal Bias in Bengali: An Evaluation of Multilingual Large Language Models Across Cultural Variations
  Azmine Toushik Wasi, Raima Islam, Mst Rafia Islam, Farig Sadeque, Taki Hasan Rafi, Dong-Kyu Chae
  WWW'25 (TheWebConf) (Short Paper) ▪ NLP ▪ LLM ▪ HAI

• Knowledge Explorer: An Agentic AI Framework for Interactive, Personalized and Multilingual Learning Experience
  Wahid Faisal*, Azmine Toushik Wasi*, Drishti Sharma*, Mahfuz Ahmed Anik, Taki Hasan Rafi, Dong-Kyu Chae
  CSCW'25 Posters → NLP ▪ HCI
  

• Exploring Bengali Creative Storytelling Capabilities of Large Language Models Across Cultural Variations
  Azmine Toushik Wasi, Raima Islam*, Mst Rafia Islam*, Farig Sadeque, Taki Hasan Rafi, Dong-Kyu Chae
  CSCW'25 Posters → NLP ▪ HCI
  

• INCLUDE: Evaluating Multilingual Language Understanding with Regional Knowledge
  Angelika Romanou, ... , Azmine Toushik Wasi, ... , Sara Hooker, Antoine Bosselut
  ICLR'25 (Spotlight, Top 5.1%) ▪ [OpenReview] ▪ Dataset ▪ NLP ▪ LLM

• GReFEL: Geometry-Aware Reliable Facial Expression Learning under Bias and Imbalanced Data Distribution
  Azmine Toushik Wasi*, Taki Hasan Rafi*, Raima Islam, Karlo Šerbetar, Dong-Kyu Chae (*Equal Contribution)
  ACCV'24 ▪ [CVF] ▪ [LNCS] ▪ [arXiv] ▪ CV

• Neural Control System for Continuous Glucose Monitoring and Maintenance
  Azmine Toushik Wasi
  ICLR'24 Tiny Papers ▪ [OpenReview] ▪ [arXiv] ▪ [GitHub] ▪ Optimization ▪ Medical AI

• When SMILES have Language: Drug Classification using Text Classification Methods on Drug SMILES Strings
  Azmine Toushik Wasi, Karlo Serbetar, Raima Islam, Taki Hasan Rafi, Dong-Kyu Chae
  ICLR'24 Tiny Papers ▪ [OpenReview] ▪ [arXiv] ▪ [GitHub] ▪ Medical AI

• Position: Adjacent Technologies Are the Key Enablers of Scalable and Safe Clinical MLLM Deployment
  Azmine Toushik Wasi, Md. Iqramul Hoque
  NeurIPS'25 → GenAI for Health and Multi-modal FMs & LLMs for Life Sciences (FM4LS) ▪ Medical AI

• Perspective: Lessons from Cybersecurity for Biological AI Safety and Regulation
  Azmine Toushik Wasi, Mst Rafia Islam
  NeurIPS'25 → Regulatable ML Workshop ▪ AI Regulation

• GFlowNets for Learning Better Drug-Drug Interaction Representations
  Azmine Toushik Wasi
  NeurIPS'25 → Structured Probabilistic Inference & Generative Modeling ▪ Drug Discovery

View All Publications and Ongoing Works →

🧑‍🔬 Research Experiences

---

• Research Collaboration, Qatar Computing Research Institute | August, 2025 - Present

• Global TA (Volunteer), How to Grow Almost Anything (HTGAA), MIT Media Lab | August, 2025 - Present

• Group Member, Schwartz Reisman Institute for Technology and Society (SRI), University of Toronto | | October 2025 – Present

• Visiting Researcher, AI4CHEMIA Research Group, King Saud University | Apr, 2025 - September, 2025

• Fellow (School of AI), Pi School | Jun, 2025 - Aug, 2025

• Global Committed Listener, How to Grow Almost Anything (HTGAA) 2025, Community Biotechnology Initiative, MIT Media Lab | Feb, 2025 - May, 2025

• Participant, Brown University Dept. of Physics presents the AI Winter School 2025 | Jan, 2025

• Student Research Assistant, Mila Quebec AI Institute | May, 2024 - March, 2025

• Community Researcher, Cohere Labs (formerly, Cohere for AI) | Aug, 2024 - October, 2025

• Founding Researcher, Computational Intelligence and Operations Laboratory, Shahjalal University of Science and Technology | March, 2021 - August, 2025

• Research Intern, Xu Lab, Carnegie Mellon University | Jul, 2024 - December, 2024

• Visiting Researcher, Data Intelligence Lab, Hanyang University | Oct, 2023 - Present

• Computational Neuroscience Summer School (Research Student), Neuromatch Academy | Jul, 2024

• CCAI Summer School (Research Student), Climate Change AI | Jun - Aug, 2024

📜 Selected Projects

---

• CIOL Presents Winter ML Bootcamp
  I am the lead organizer and instructor of the ML bootcamp, with more than 50 participants. I prepared materials and took live sessions on EDA, Tabular Data Modeling, Hyperparameter Tuning, ANNs - Deep Learning, NLP, Computer Vision, LLM Agents and mentored 6 research teams in the bootcamp.
  [Website] ▪ [GitHub] (Check Session Notebboks) ▪ [Youtube] (Check Recordings)
• AyaFestPe: A multi-lingual and multi-cultural festival exploration guide
  The pipeline starts by collecting festival data from HF and translating it with AyaExpanse for multilingual access. We then gather user queries, use M-RAG to embed data and queries, rerank results, and combine relevant context with the query to generate output.
  [GitHub] ▪ [Colab] (using Cohere API) ▪ [Kaggle] (Using released model weights, with quantization)
• 🏫 Online ML University: Free AI resource collection for everyone! Currently ~150⭐ in GitHub!
• 📑 Paper Organization and Collection
  
  • ICML 2024 Graph Paper Collection : List of all graph and/or GNN papers accepted at ICML'24. Currently ~160⭐ in GitHub!
  • ICLR 2024 Graph Paper Collection : List of all graph papers accepted at ICLR'24. Currently ~70⭐ in GitHub!
  • ICLR 2024 LLM Paper Collection : List of all LLM papers accepted at ICLR'24. Currently ~35⭐ in GitHub!

👩‍💻 Technical Skills

---

• Programming Languages: Python (Advanced), C, C++, MATLAB, R, SQL
• DS & ML Tools (Python): NumPy, Pandas, Matplotlib, Seaborn, Scikit-learn, TensorFlow, PyTorch, PyCaret, LangChain, VLLM, Pydantic
• Data Science Techniques: EDA, Experimental Design, Hypothesis Testing, Sampling, Statistical Inference, Data-driven Decision Making
• Machine Learning Techniques: SStatistical ML Methods, Deep Learning, NLP, Computer Vision, Graph Neural Networks (GNNs), GFlowNets, Flow Matching, Diffusion Models, RL and Reasoning in LLMs, Self-Verification, Uncertainty, Agentic Decision-Making, AI Reasoning, RAG, AI Agent Environments and Systems and Reward-Based RL Fine-Tuning
• Biomedical AI & Clinical Applications: Molecular Properties, Interaction, Dynamics, Optimization and Simulation; De Novo Molecule Design/Dynamics with GNNs, RL/Reward/Energy-Guided Modeling, Flow Matching, Diffusion; AI-based Drug Discovery and Genomics; Single Cells and Multi-Omics and AI; Knowledge Graphs in Biomedicine: Bio/Medical/Clinical Agents and NLP Applications; Lab Automation (Opentrons and Ginkgo); Analysis and Modeling with Benchling
• AI Safety, Governance, and Responsible Development: Alignment, value-sensitive design, interpretability, risk assessment, governance frameworks, policy and regulatory compliance, auditability, model monitoring, and responsible deployment.
• Data Analysis & Visualization: MS Excel, Power BI, Tableau, SAS
• Interdisciplinary Research: AI for Good, multilinguality and multicultural alignment, accessibility design in low resource settings, fairness and unbiased representation in clinical and social contexts, human factors and ergonomics evaluation, integrating local knowledge and cultural values into AI systems, and sociotechnical risk mitigation for mental health and clinical deployments.
• Human Computer Interaction (HCI): Clinical and patient centric AI workflows, LLM customization and prompt interface design, scalable survey and user study pipelines, data collection and annotation with domain experts, UI and UX frameworks for trust and comprehension, participatory design with healthcare stakeholders, and evaluation methods that capture cognitive load, usability, safety, and user experience in high risk environments.
• Automation & Productivity: MS Word, PowerPoint, Excel Automation; Google Sheets Scripting; Python-based Automation; Adobe Photoshop and Illustrator
• Other Tools & Skills: GitHub, VS Code, Azure, AnyScale, Replit, Colab, Kaggle, Parallel & Distributed Computing, Product & Project Management, Strategic Planning

💼 Academic Services

---

• Conference or Journal Reviewer
  • AI/ML: ICLR, NeurIPS
  • NLP: ACL ARR Feb (ACL, EMNLP, NAACL, COLING) [View in official reviewer list]; COLING.
  • HCI/HAI: CHI, CSCW, UbiComp/ISW, TEI, ICM, IDC.
  • Journals: Transactions on Pattern Analysis and Machine Intelligence (T-PAMI) (IF 18.6), IEEE Transactions on Affective Computing (TAFFC) (IF 9.8), Complex & Intelligent Systems (Springer Nature) (IF 5.6).
  • Total conference or journal paper reviewed: 87
• Workshop Program Committee and Reviewer
  • NeurIPS'25: AI4Science , AI4Materials, Scaling Environments for Agents, Structured Probabilistic Inference & Generative Modeling, Regulatable ML, Frontiers in Probabilistic Inference, MusiML , EurIPS: Multimodal Representation Learning for Healthcare
  • EMNLP'25: Multulingual Represent Learning
  • ICLR'25: AI4MaterialsDiscovery Re-Align, LMRL
  • NAACL'25: DravidianLanTech, CLPsych, In2Writing
  • AAAI'25: Social Impact of AI,
  • CHI'25: STAIG, HEAL
  • ICML'25: World Models, MusIML
  • NeurIPS'24: Behavioral ML, AI4MaterialsDiscovery, EvalEval, Military AI, MusIML
  • EMNLP'24: NLP for Positive Impact, Multulingual Represent Learning
  • ICML'24: AI4Science, LLMs & Cognition
  • ACL'24: TextGraphs-17, ClimateNLP, SMM4H, Language + Molecules, GenderBiasNLP, WASSA.
  • MICCAI'24: Fairness in Medical Imaging, GRaphs in biomedicAl Image anaLysis.
  • Total top conference workshop paper reviewed: 90
• Organizer
  • NeurIPS'24, ICML'24, NeurIPS'25 Muslims in ML Affinity Workshop (Steering Committee Member)

Copyright © 2025 Azmine Toushik Wasi. All rights reserved.