Sankalp Pandey

University of Arkansas

Ph.D. in Computer Science, advised by Dr. Khoa Luu

Supported by Distinguished Doctoral Fellowship (~$45,000/year total funding for 4 years)

University of Arkansas

B.S. in Computer Science & B.S. in Computer Engineering

Minors: Mathematics, Data Analytics

Senior of Significance

Supported by:

• Arkansas Governor's Distinguished Scholarship ($10,000/year for 4 years)
• University of Arkansas Chancellor's Scholarship ($8,000/year for 4 years)
• University of Arkansas Honors College Research Grant ($4,000)
• Farison Scholarship, College of Engineering ($1,000)
• University of Arkansas Honors College Conference Travel Grant ($1,000)

Agentic Multi-Turn Reasoning: A Fairness Approach

Thanh-Dat Truong, Sankalp Pandey, Hugh Churchill, Jackson Cothren, Marios Savvides, Khoa Luu

Fair Multi-Level Preference Optimization (Φ-MPO) for long-horizon credit assignment and imbalanced data in multi-turn agents.

Project, PDF, arXiv

QuPAINT: Physics-Aware Instruction Tuning Approach to Quantum Material Discovery

Xuan-Bac Nguyen, Sankalp Pandey, Suhyun Kim, Hojin Jang, Arabinda Kumar Choudhary, Khoa Luu

A physics-aware instruction-tuning pipeline that improves multimodal reasoning for quantum material discovery.

Project, PDF, arXiv, Code

CLIFF: Continual Learning for Incremental Flake Features in 2D Material Identification

Sankalp Pandey, Xuan-Bac Nguyen, Nicholas Borys, Hugh Churchill, Khoa Luu

A continual learning strategy for incremental flake-feature identification with reduced catastrophic forgetting.

PDF, arXiv, Code

QMoE: A Quantum Mixture of Experts Framework for Scalable Quantum Neural Networks

Hoang-Quan Nguyen, Xuan-Bac Nguyen, Sankalp Pandey, Samee U. Khan, Ilya Safro, Khoa Luu

A quantum mixture-of-experts architecture that routes inputs to specialized experts for scalable quantum neural networks.

PDF, arXiv

Hidalga Technologies

Machine Learning Engineer Intern. Built healthcare AI pipelines for prior-authorization prediction, synthetic FHIR simulation, and automated document validation.

Algonauts 2025 Vision-Brain Encoding Challenge

Ranked 5th out of 67 teams in the 2025 international vision-brain encoding competition.