I build AI systems that are interpretable, accountable, and trustworthy. My work sits at the intersection of machine learning, NLP, and human-centered computing across health, misinformation, and responsible AI.
AI systems increasingly shape high-stakes decisions, from healthcare to information access to opportunity allocation. Yet the people most affected often cannot understand or contest what these systems produce. That gap drives my research.
I work at the intersection of AI and human-centered computing, combining technical ML work with empirical studies of how people experience AI in practice. On the technical side, I have built NLP classifiers for misinformation detection [1], trained models on clinical data and developed physician-validated fuzzy logic systems for stroke risk prediction [2]. On the human side, I conduct cross-cultural qualitative interviews across regions [3], run participatory studies with clinicians, developers, and end users [4], and use mixed methods to study real-world interaction with AI [5]. My recent work produced a participatory five-pillar explainability framework for AI-driven health applications, published at JMIR (Q1, IF 6.0). My research has been published in JMIR, ACM CHI, ACM UbiComp, ACM JCSS, IEEE, Springer, and Oxford University Press.
My work is grounded in a simple conviction: meaningful AI systems cannot be designed without understanding how people think, and human-AI interaction cannot be studied without understanding what the system is actually doing.
I also teach. I serve as an Adjunct Lecturer at Independent University, Bangladesh and Senior Lab Instructor at North South University, where I have worked with over 1,000 undergrad students across programming and computing courses.
I completed my MSc in Computer Science at Independent University, Bangladesh, and my BSc in Computer Science and Engineering at North South University. I am currently seeking PhD opportunities to pursue rigorous research at the intersection of machine learning and human-centered computing, where technical work and empirical understanding of people inform each other.
"My goal is to make AI systems that are not just accurate, but interpretable, accountable, and genuinely useful for the people who need them most."
I am currently auditing bias in LLM generated clinical recommendations, studying how explanations shape human trust, and grounding everything in real patient data. As this work matures, I intend to push deeper into model evaluation, bias mitigation, and explanation generation, toward clinical AI that is not just capable, but fair and trustworthy.
Related work 1: Who Gets What Advice? Counterfactual and Linguistic Audits of Bias in LLM Clinical Recommendations — Under Submission
Related work 2: The Explainability Illusion: Formalising Localisation as a Necessary Condition for XAI in Health AI — Under Submission
Related work 3: When AI Explains Itself: Do LLM Explanations Help or Mislead Human Decision-Making? — In Progress
Happy to discuss PhD opportunities, research collaborations, or speaking invitations. I'll respond within a few working days.
