Arpan Poudel’s Website

I am a Ph.D. student in Computer Science at the University of Arkansas under Dr. Qinghua Li. My work is in federated learning, large language models and medical imaging, with a focus on building reliable and privacy-preserving AI systems for healthcare.

Research interests

• Vision–language models
• Generative models
• Image reconstruction
• Federated learning

Current and recent projects

• Federated vision–language modeling for chest X-rays. I fine-tune Qwen-VL models with LoRA adapters and explore router/gating methods to choose adapters across clients without sharing data. I use structured report templates (e.g., Findings / Impression) to reduce hallucinations and improve clinical usefulness.

• MRI super-resolution. I developed an unsupervised diffusion approach that enhances low-resolution MRI without paired training data. I also released the SRMRI dataset with matched low-/high-resolution scans to enable fair benchmarking and to study scan-time reduction via super-resolution.

Outside research

I enjoy cricket, video games, fishing, and cooking.

Collaboration

I am open to collaborations in computer vision and medical imaging—especially projects that connect rigorous methods with real clinical needs. email: arpanpoudel@gmail.com