Recent Advances in Bangladesh’s Computer Science Research

Shaikh Afnan Birahim

In recent years, Bangladesh has emerged as a vibrant centre for computer science innovation, producing cutting-edge research and leveraging artificial intelligence to advance healthcare, legal access, and public safety. One of the most striking contributions is a study titled “Transfer Learning and Explainable AI for Brain Tumour Classification: A Study Using MRI Data from Bangladesh”, published in June 2025. This work tackles a pressing health challenge in resource-constrained settings: the early detection of brain tumours through automated MRI analysis.

By fine-tuning pre-trained models such as VGG16, VGG19, and ResNet50 on local datasets, the researchers achieved remarkable accuracy. VGG16 reached 99.17% in distinguishing between glioma, meningioma and other tumours. More importantly, the study integrated Explainable AI techniques, such as Grad-CAM and Grad-CAM++, offering visual maps that reveal which regions of the MRI the model focused on. This transparency is crucial for clinical adoption in Bangladesh, where trust in automated systems relies on their interpretability.

Equally notable is the pioneering framework LegalRAG: A Hybrid RAG System for Multilingual Legal Information Retrieval. Bangladesh’s legal texts are often complex and span both Bangla and English, complicating access for non-specialists. LegalRAG bridges this divide by combining retrieval-augmented generation (RAG) with tailored indexing strategies, enabling users to ask legal questions and receive precise, bilingual answers.

Evaluated on the Bangladesh Police Gazettes, it significantly outperforms existing baselines across multiple retrieval metrics. The implications are profound: greater legal literacy for underserved communities and potential integration into government and NGO platforms to streamline legal aid in rural areas. This project showcases not only technical sophistication but also profound social relevance.

In the realm of public safety, a recent (April 2025) study titled ‘Design and Application of Multimodal Large Language Model Based System for End to End Automation of Accident Dataset Generation’ demonstrates how LLMs and web scraping can transform data collection. The system combines automated code generation, accident news classification, structured data extraction, and de-duplication to harvest and process reports from 14 major Bangladeshi news outlets. Over a span of 111 days, from October 2024 to January 2025, the system processed more than 15,000 articles and identified 705 unique accident cases.

Evaluation of the code generation module yielded an impressive 91.3% calibration accuracy and 80% validation rate. Geographic breakdowns revealed that Chittagong and Dhaka accounted for the majority of incidents. By delivering a reliable, scalable and transparent accident database, this work lays the groundwork for data-driven road safety policies and resource allocation.

What binds these breakthroughs is a shared methodology: transfer learning, Explainable AI, hybrid retrieval architectures, and innovative use of LLMs, all applied thoughtfully to context-specific challenges. In the healthcare domain, deep learning models trained on international architecture are fine-tuned with local Bengali datasets and augmented with interpretability to meet clinical standards. In legal services, RAG systems integrate bilingual documents, addressing both linguistic diversity and societal barriers. Meanwhile, in accident data collection, multimodal LLM pipelines automate tedious processes, democratising data access for policymakers and the public.

Bangladesh’s computer science community is also organising major forums as well as numerous IEEE conferences. These events foster collaboration and visibility for home-grown research groups across universities like BUET, DUET and SUST, and help translate prototypes into deployment-ready solutions.
Taken together, these studies reflect a dynamic research ecosystem in Bangladesh, one that is pragmatic, socially responsive and globally aware. By linking advanced algorithms with deep contextual understanding, Bangladeshi computer scientists are poised to make meaningful contributions that resonate far beyond academic circles. Their work not only exemplifies technical excellence but also demonstrates a clear commitment to addressing local needs, from saving lives and ensuring justice to improving public welfare through informed policy. In short, Bangladesh is positioning itself as an AI innovation hub in South Asia, offering scalable, impactful solutions to the challenges of our time.
____________________________________
The writer is a postgraduate student of Computing Science at The University 
of Glasgow