Research Assistant, Machine Intelligence Lab
North South University · Dhaka, Bangladesh
I work on multimodal learning, retrieval, and speech-driven visual systems. At MILab, I am advised by Dr. Nabeel Mohammed and Dr. Shafin Rahman. My work includes multimodal fusion, low-resource language understanding, and the engineering of conversational avatars.
My current research examines evidence loss in multi-hop question answering (ERL), the relationship between speech and facial motion (RIFT), and capacity growth in verified language-model self-improvement (RSI).
Research interests. Multimodal and representation learning; memory and retrieval; long-video understanding; 3D vision and neural rendering. I am also interested in predictive representations and world models, as exploratory directions.
I am applying for Fall 2027 Ph.D. positions and also considering thesis-based M.Sc. opportunities.
CAT-GS accepted; now published in Neurocomputing. Our paper on balanced multimodal learning was accepted for publication and is now available from the publisher.
New research direction. Started RSI, an exploratory study of capacity growth in verified language-model self-improvement.
AACL-IJCNLP acceptance. Can LLMs Follow the Pulse of a Crisis? was accepted to the 2026 Main Conference. The paper is available on OpenReview.
CAT-GS preprint. The public manuscript became available.
Tuctuc role concluded. Completed my Junior Software Engineer role, working remotely on Odoo-based ERP systems and React Native applications.
PLOS ONE publication. Our rainfall-classification study was published in PLOS ONE.
PLOS ONE acceptance. Ensemble and Temporal Feature-Based Framework for Rainfall Classification in Bangladesh was accepted for publication.
Completed B.Sc. requirements. Computer Science and Engineering at North South University; CGPA 3.80/4.00, Summa Cum Laude.
PNEUMA thesis defense. Defended our four-person undergraduate thesis on long-video memory under Dr. Nabeel Mohammed. Read about PNEUMA’s memory architecture and open questions.
IEEE ICCIT 2025. When a Nation Speaks, our study of sentiment during Bangladesh's 2024 mass uprising, was a conference paper at ICCIT 2025.
Joined MILab. Began as a Research Assistant at the Machine Intelligence Lab, North South University, advised by Dr. Nabeel Mohammed and Dr. Shafin Rahman.
Joined Tuctuc. Started as a remote Junior Software Engineer.
Published and accepted
Neurocomputing, article 135392 (2026) · Published
I contributed comparative evaluation protocols, baseline analysis across audio–visual benchmarks, and pipeline-level optimization experiments.
AACL-IJCNLP 2026, Main Conference · Accepted · Paper available
Paper · OpenReviewarXivDataset page
I contributed corpus phase partitioning, LoRA/SFT experiments, the parent-post context ablation, and topic and temporal analyses.
PLOS ONE 21(3): e0342646 · Published
I contributed to conceptualization, temporal-feature methodology, and manuscript review and editing.
2025 28th International Conference on Computer and Information Technology (ICCIT), IEEE, pp. 413–418 · Published
I contributed data collection and filtering, LDA topic modeling, baseline evaluation, and dataset and correlation figures.
I study how candidate access, evidence selection, and reading affect multi-hop answers. Preliminary findings motivate further comparisons with stronger retrieval and scoring methods.
I investigate how speech-predictable facial motion relates to rendered articulation. Following a negative predictive pilot, RIFT-M focuses on measurement reliability.
I am building a verified self-training pilot to study model expansion. Reliable training targets and recovery checks come before testing the growth hypothesis.
Collaborative MILab and industry work on speech-to-avatar integration, distributed inference, and Jetson deployment. Niloy and I are exploring a manuscript direction from the deployed PersonaPlex–IMTalker pipeline.
All research and projects, including PNEUMA, AirFLOW, DynApex, and the StreamTalker-GS negative-result study.
North South University · March 2025–present
Research in multimodal learning, low-resource NLP, and speech-driven visual systems. My collaborative work includes distributed inference, avatar integration, and evaluation under deployment constraints.
North South University · 2021–2025; requirements completed December 2025
Summa Cum Laude · CGPA 3.80/4.00
Spain, remote · January 2022–June 2026
Odoo-based ERP systems and cross-platform mobile applications.
I participated in the ICPC Asia Dhaka Preliminary Contest in 2021 and received an NSU Academic Scholarship in 2023. Full CV, coursework, and technical background.
For research discussions, collaboration, or graduate supervision, please write to tanvirahmedkhan74@gmail.com.