Adapting to change
Robust Policy Optimization
RPO is integrated into CleanRL and skrl, and used in Physical Intelligence’s released implementation of real-time action chunking.
Method and adoption →
Postdoctoral Researcher · Purdue University
Md Masudur Rahman
Learning-based autonomy for surgery and other high-stakes settings.
I develop reinforcement learning and foundation-model methods for systems that act reliably when conditions change and recognize when to hand control back to a person, with surgical and medical robotics as the driving application.
I am a postdoctoral researcher in Purdue’s Edwardson School of Industrial Engineering, working with Juan P. Wachs. I received my Ph.D. in Computer Science from Purdue in 2024, advised by Yexiang Xue.
Research agenda
When should an autonomous system act, and when should it hand control to a person?
In surgery, capability alone is not enough: a system must also recognize when a situation has moved beyond what it can handle. My research builds toward that in three connected parts: learning methods that stay reliable when conditions change, multimodal reasoning that grounds decisions in clinical evidence, and shared-control designs that divide work between the system and the clinician. Each is evaluated against the disruptions and failures that occur in real procedures.
The same questions apply to robot learning and decision systems beyond surgery, wherever errors are costly.
Explore my research program →Funding & recognition
Full funding record →- NSF EAGERKey personnel · Foundation models in reinforcement learning · 2025–2027
- NAIRR compute allocationPrincipal investigator · Argonne Cerebras CS-3 · 2026
- Clinical pilot fundingTeam lead · Purdue CTSI and Health of the Forces · 2025–2027
- Best Paper AwardMIUA 2025
- Oral presentationAAAI/IAAI 2026 · 15% acceptance
- Gold ReviewerICML 2026
Selected contributions
All publications →Reasoning from evidence
AI for burn diagnosis
AURA uses vision-language models to combine visual evidence with clinical knowledge for burn assessment. AMBUSH-AI, a non-invasive burn diagnostic system, appears in Annals of Surgery.
IAAI 2026 oral · Annals of Surgery 2026
Research and papers →Sharing control
Semi-autonomous surgical teleoperation
SARTRES is a semi-autonomous teleoperation environment for surgery, studying how learned robot skills can be combined with surgeon guidance.
CMBBE: Imaging & Visualization 2021
Research and paper →Teaching & mentoring
Teaching across computing, machine learning, and medical AI.
I taught Data Structures and Digital Logic Design as a Lecturer at BRAC University, have given guest lectures in Purdue’s IE 549: Machine Vision in Intelligent Robotic Systems, and am co-teaching a course on AI in medicine and large language models. I currently mentor three Ph.D. students in Industrial and Biomedical Engineering on medical AI, reinforcement learning with foundation models, and medical robotics.
Teaching and mentoring record →Selected updates · 2025–2026
News & highlights
Research support. NAIRR Compute Support as Principal Investigator, with an allocation on Argonne’s Cerebras CS-3 system. Details →
New publication. “A Multimodal Artificial Intelligence Reasoning Framework for Burn Diagnosis” published in Military Medicine. Paper ↗
Research service. Recognized as a Gold Reviewer at ICML 2026. Recognition →
Paper acceptance. AMBUSH-AI, our work on automated non-invasive burn diagnosis, accepted at Annals of Surgery. Publication details →
Paper acceptance. AURA accepted for an oral presentation at IAAI 2026. Research and paper →
Best Paper Award. Recognized at MIUA 2025 for work on knowledge-driven burn diagnosis with vision-language models. Purdue feature ↗