Portrait of Sumaiya Karim Katha

Sumaiya Karim Katha

About

I study whether the attention of a large language model is a faithful explanation of how it reaches its answer. An explanation can look convincing and still be unfaithful, and the people who rely on it can be misled.

I also build LLM agents. I am working on an agentic system that checks the document packets people submit for official applications and finds missing or wrong documents. With Dr. Amin Ahsan Ali, I am co-designing LangTutor, an AI tutor that adapts each lesson to the learner's level and past mistakes. For Kotoba, my online shop, I built a multi-agent system that answers customer messages, places orders and hands harder cases to a person.

Publications (3)

Abu Tyeb Azad, Ishita Sur Apan, Fahim Ahmed, Sumaiya Karim Katha, Ezharuddin Jubaer, Armun Alam, Pranjal Kumar Nandi, Amin Ahsan Ali, Aman Chadha, Md Mofijul Islam, AKM Mahbubur Rahman

BaFCo: A Document Understanding Benchmark for Complex Bangla Form Comprehension

Computer Vision – ECCV 2026, pp. 402-420, 2026

Book ChapterDOIURL
  • key information extraction
  • multimodal large language models
  • document layout analysis
  • government forms
  • chain-of-thought prompting
  • low-resource languages
  • zero-shot prompting
  • document understanding
  • Bengali language
  • annotated dataset

Sami Rashid, Md Fahim Shahriar, Sumaiya Karim Katha, Ezharuddin Jubaer, Akm Moshiur Rahman Mazumder, Ashraful Islam, Mohsena Ashraf, Asif Mahmud, M Ashraful Amin

AI Agents in Healthcare: A Survey of Applications, Resources, and Challenges

2025 IEEE 23rd Student Conference on Research and Development (SCOReD), pp. 1-6, 2025

Conference PaperDOIURL
  • algorithmic bias
  • Medical education
  • diagnostic imaging
  • clinical decision support
  • multi-agent systems
  • medical consultation
  • large language models
  • AI Agents
  • regulatory challenges
  • explainability
  • SDG 3Good Health and Well-being

Projects (4)