Computer Science and Engineering undergraduate at BUET
Machine Learning · Computer Vision · Speech
I am a B.Sc. student in Computer Science and Engineering at the Bangladesh University of Engineering and Technology (BUET). My work centers on systematic evaluation of machine-learning systems, with particular interest in robustness, representation learning, and model behavior under distribution and environmental change.
- Computer vision and visual place recognition
- Image retrieval and visual embeddings
- Vision-language and multimodal learning
- Robust and trustworthy machine learning
- Bengali speech recognition and speaker diarization
-
From Image Resizing to Adaptive Pipeline: Improving Visual Place Recognition Without Re-Training
A H M Fuad, Risalat Labib
ACM NSysS 2025, pp. 142–147. DOI
A study of test-time resolution, augmentation, and environmental conditions across pretrained VPR methods, including a condition-aware method-selection pipeline. Experiments were built on the public VPR Methods Evaluation framework. -
Make It Hard to Hear, Easy to Learn: Long-Form Bengali ASR and Speaker Diarization via Extreme Augmentation and Perfect Alignment
Sanjid Hasan, Risalat Labib, A H M Fuad, Bayazid Hasan
arXiv, 2026. Paper
Long-form Bengali ASR and speaker-diarization experiments associated with Lipi-Ghor-882 and Team Villagers' DL Sprint 4.0 submission.
- Test-time optimization for visual place recognition — Experiment artifacts for resolution sweeps, database augmentation, and condition-aware evaluation of pretrained VPR methods on AmsterTime, SPED, and SVOX. Built on and clearly attributed to the public VPR Methods Evaluation framework. Repository
- Bengali ASR and speaker diarization — Experiment notebooks for model evaluation, fine-tuning, acoustic augmentation, and diarization from BUET CSE Fest DL Sprint 4.0. Team Villagers placed 2nd Runner-Up and received the Best Dataset Award. Repository
- Exoplanet candidate classification — A Streamlit application and stacked tabular-ML pipeline for classifying Kepler Objects of Interest, developed for the NASA Space Apps Challenge. Repository
- 2nd Runner-Up and Best Dataset Award — BUET CSE Fest DL Sprint 4.0, Team Villagers
- 6th place, private leaderboard — IUT Datathon “Olikbochon,” Team IUFree