Overview
Aspiring researcher with a strong and evolving foundation in deep learning, computer vision, and model efficiency, built through hands-on experience with fine-grained classification, vision transformers, and contrastive learning. Demonstrated ability to design and implement reproducible, end-to-end research pipelines and build state-of-the-art architectures from scratch. Passionate about driving innovation in biomedical image analysis, signal processing, and computer vision through focused empirical research and impactful open-source collaboration.
Projects
VisionDesk
Real-time employee monitoring using YOLOv8, FastAPI, and manual annotations. [Code]
Vision Transformer (ViT)
ViT from scratch with patch embedding, attention visualization. [Code]
Swin Transformer
Implemented window-based attention with shift and benchmarked classification gains.
CutMix & MixUp
Data augmentation via mixing inputs/labels in PyTorch. [Code]
Knowledge Distillation
ResNet50 → MobileNetV3 on STL-10 using tuned soft-label training. [Code]
SimCLR
Self-supervised contrastive learning with projection heads. [Code]
U-Net Segmentation
Segmentation with U-Net/ResNet encoders, modular design. [Code]
Conditional GAN
MNIST digit generation using label-conditional GAN in Keras. [Code]
Flower Classification
CNNs and transfer learning on flower species dataset. [Code]
Flood Segmentation
Segmented satellite flood imagery using U-Net. [Code]
Education
- B.Sc. in CSE – University of Rajshahi (2024), CGPA: 3.90/4.00
- Higher Secondary Certificate – Rajshahi College (2018), GPA: 5.00/5.00
- Secondary School Certificate – Govt. Laboratory High School (2016), GPA: 5.00/5.00
Skills
- Deep Learning, Vision Transformers
- Contrastive Learning, Signal Processing
- Python, C/C++, Java, JavaScript
- PyTorch, TensorFlow, Huggingface
- Flask, FastAPI, REST APIs
- MySQL, MongoDB, SQLite
- OpenCV, NumPy, Pandas, scikit-learn
- Git, GitHub, Jupyter, LaTeX, Linux
Publications
[J.1] Sangeeta Biswas, Md. Ahanaf Arif Khan, Md. Hasnain Ali, Johan Rohdin, Subrata Pramanik, Md. Iqbal Aziz Khan, Sanjoy Kumar Chakravarty, Bimal Kumar Pramanik (2025). "Interpreting Deep Neural Networks in Diabetic Retinopathy Grading: A Comparison with Human Decision Criteria". Life, 15(9), 1473. DOI
[C.1] Md. Ahanaf Arif Khan, Md. Hasnain Ali, Nirjor Saha, Md. Sadman Shakib Shoumik, Sangeeta Biswas (2023). "Competency Comparison of Deep Neural Networks for Identifying Gender in Color Fundus Photographs". In 26th International Conference on Computer and Information Technology (ICCIT), IEEE. DOI
Awards & Achievements
UGC Stipend 2025: Highest academic distinction in Faculty of Engineering
AI Hackathon Champion (2025): VisionDesk developed during manufacturing track
Robi Datathon Finalist (2024): Top 7 out of 1000 teams in ML business challenge
Dean's Award (2023): Twice recognized for academic performance
Seminars & Workshops
- Bangladesh AI Summit (2025): Hosted by Bangladesh Innovation Conclave
- IC4ME2 Poster Presentations (2021 & 2022): Presented work at University of Rajshahi
Certifications
- Python Data Associate (Feb 2025): Issued by DataCamp [Verify]
Additional Information
- Languages: Bengali (Native), English (Professional Proficiency)
- Interests: Traveling, Watching Movies, Exploring new ideas and technologies