AI Researcher · Legal NLP · Low-Resource AI
I build reliable AI systems for difficult language, legal, and resource-constrained problems where data, supervision, and compute are limited.
Hello, I’m M. Mohaiminul Islam, an AI researcher working at the intersection of language, structured knowledge, and reliable machine learning. I completed my Bachelor of Engineering in Computer Science and Telecommunication Engineering at Noakhali Science and Technology University and my Master of Science in Computer Science at South Asian University, New Delhi.
My research focuses on legal text analytics, low-resource language technologies, natural language processing, and speech processing. In legal AI, I study extractive question answering, information retrieval, retrieval-augmented generation, knowledge graphs, and graph-based reasoning for Indian intellectual property law. In language technology, I work on Bengali regional dialects and code-switched Bengali-English speech, with an emphasis on datasets, robust evaluation, and systems that remain useful when data and supervision are limited.
I am also interested in large language models, computer vision, and Efficient & Lightweight AI. In particular, I am interested in medical image segmentation and in building efficient models that can deliver accurate, reliable results with limited computational resources. My goal is to design practical models that balance segmentation quality, inference speed, memory use, and real-world deployment constraints. Across these areas, I aim to develop AI systems that are more factual, explainable, and dependable.
Beyond research, I enjoy history, philosophy, theology, and literature. I have also translated The Lessons of History by Ariel Durant and Will Durant, an experience that reflects my long-standing interest in ideas, language, and the way knowledge travels across cultures.
CGPA: 3.50 out of 4.00
CGPA: 3.14 out of 4.00
GPA: 5.00 out of 5.00
GPA: 5.00 out of 5.00
Publications & Research
My work connects legal text analytics, information retrieval, knowledge graphs, low-resource speech, and reliable machine learning. The status labels below reflect the current submission record.
IEEE DSAA 2026 - 13th IEEE International Conference on Data Science and Advanced Analytics (Core B)
Accepted to IEEE DSAA 2026 for Long (Oral) Presentation. The conference will be held in New Delhi, India, from 6-9 October 2026. Submission #147.
Submitted to a CORE B-ranked conference
A validity-focused approach to region-based parallel mining that preserves sender locality while optimising semantic similarity, interaction cuts, and load under capacity constraints.
Language Resources and Evaluation (LRE)
A 41,499-clip benchmark covering 15 Bangladeshi regional dialects and naturally occurring Bengali-English code-switching, with an automated processing and quality-filtering pipeline.
Artificial Intelligence and Law
A diagnostic study of retrieval failure, statutory citation resolution, and domain-adaptive pretraining for extractive QA over Indian intellectual property statutes.
Artificial Intelligence and Law
An audit of ontology conformance, baseline integrity, benchmark headroom, and the actual contribution of knowledge-graph evidence to legal QA.
Frontiers in Communication
A review of Bengali dialect ASR, self-supervised speech encoders, multilingual models, data scarcity, noise, and future directions for practical low-resource speech systems.
Database Systems, Data Mining & Network Science
Worked with MySQL, PL/SQL, and MongoDB, covering schema design, normalization, query optimization, triggers, and aggregation pipelines.
Taught preprocessing, classification with SVM, k-NN and decision trees, clustering with K-Means and DBSCAN, text mining, web mining, and PCA.
Explained centrality, community detection, network partitioning, DAGs, and shortest paths using Neo4j and Cypher.
Research Scholar
Data Scientist
Data Engineer
Blockchain Researcher
Research Assistant
New Delhi , India
+8801787619730
