Md Naim Hassan Saykat

GitHub · LinkedIn · ORCID · Email

I am an artificial intelligence researcher working on medical image analysis, computer vision and explainable AI, and on AI governance for the public sector. I am a Master’s student in Artificial Intelligence at Université Paris-Saclay and an independent researcher based in Bangladesh.

My research in medical imaging focuses on models that remain reliable beyond the data they were trained on: external evaluation on independent datasets, calibration and uncertainty, and visual explanations that clinicians can inspect. In parallel, I lead the Bangladesh AI Policy & Innovation Lab (BAPIL), an independent research initiative that designs how AI can support evidence-based policymaking in Bangladesh safely and openly.

Research interests

• Medical image analysis, in particular dermoscopic skin-lesion classification
• Cross-dataset generalization and external evaluation (for example, HAM10000 to ISIC 2019)
• Explainable and trustworthy AI: Grad-CAM, calibration, uncertainty estimation
• Vision Transformers and CNN architectures, and efficient models for resource-constrained settings
• AI governance and policy intelligence: responsible AI, risk assessment, bilingual (Bangla and English) information retrieval for public policy

Selected work

• Generalizable skin-lesion classification: seven deep learning architectures and a probability-level ensemble, evaluated internally on HAM10000 and externally on ISIC 2019, with calibration, bootstrap confidence intervals and Grad-CAM. Revised manuscript under review. Project · Code · DOI
• BAPIL Technical Blueprint (2026): a buildable plan, in English and Bangla, for an AI policy-intelligence capability in Bangladesh, with an institutional design, reference architecture, governance and assurance system, tested reference implementation and 36-month delivery plan. Project · DOI
• Information retrieval, generative models and NLP: patent retrieval with cross-encoder re-ranking, CycleGAN image translation, a DRAW-style recurrent VAE and text emotion recognition. All projects

Opportunities

I am open to research collaborations and PhD positions in medical image analysis, trustworthy AI, or AI for public policy. Feel free to contact me by email.