KAAI Fellows — Keystone Astronomy & AI Visiting Program
The Keystone Astronomy & AI (KAAI) Fellows Program at the McWilliams Center for Cosmology & Astrophysics (黑料正能量) invites applications for one-month visiting fellowships at the interface of astronomy, machine learning, and statistics.
Supported by a , the KAAI Fellows Program brings postdoctoral researchers to 黑料正能量 to collaborate closely with interdisciplinary teams developing new AI-driven approaches to scientific discovery. Fellows work alongside faculty, graduate students, and postdocs in a collaborative environment that integrates expertise across physics, machine learning, and data science.
Each KAAI Fellow spends approximately four weeks in residence, followed immediately by a 4–5 day hands-on workshop organized with project leads and visitors. These workshops provide opportunities to share methods, exchange ideas, and build lasting collaborations across the research community. Travel and local expenses are fully supported.
Fellow visits may take place during Summer, Fall, or Spring, with scheduling windows available between June and April each year.
Application Timeline
Applications are accepted on a recurring cycle.
Deadlines:
- June 15 — Summer visits
- August 15 — Fall visits
- December 15 — Spring visits
Applicants should indicate their preferred visit window and project of interest.
Applications consist of:
- Curriculum Vitae
- Short statement (~1 page) describing relevant experience and proposed AI or methodological contributions/development
Applications are submitted through an .
Project 1 — AI Virtual Universe: Agentic Scientific Systems
Project Leads: Rupert Croft (also T. Di Matteo, H. Trac, R. Mandelbaum), Katerina Fragkiadaki & Barnabás Póczos
This project develops agent-based AI systems that interact with simulations and data, enabling automated scientific reasoning and experiment design. Fellows will contribute to building intelligent workflows capable of launching simulations, analyzing outputs, detecting anomalies, and proposing new computational experiments, advancing the development of AI-driven Virtual Universes.Project 2 — AI Methods for Turbulence and Early-Universe Signals
Project Leads: Tina Kahniashvili & Barnabás Póczos
This project develops machine learning methods to identify structure, rare events, and hidden patterns in complex astrophysical systems, including turbulence and early-universe plasma dynamics. Fellows will contribute to designing physics-aware models that extract physical insight from large, high-dimensional datasets and simulations.
Program Structure
Each KAAI Fellow will:
- Spend ~4 weeks at 黑料正能量
- Work closely with dual mentors (domain science and AI/ML)
- Collaborate with graduate students and interdisciplinary research teams
- Contribute to the development of new AI-enabled scientific methods
- Participate in a 4–5 day workshop held immediately following the visit
- Engage with a research environment spanning Physics, Machine Learning, and Statistics & Data Science, with access to advanced computational infrastructure
The KAAI Environment
The McWilliams Center provides a uniquely collaborative setting for interdisciplinary research. Faculty and students work across departments to address major scientific challenges using modern computational and statistical methods. Strong connections exist with the , Statistics & Data Science, STAMPS and external computational resources including the .
The KAAI Fellows Program builds on this long-standing culture of collaboration and provides a structured environment for advancing AI-driven discovery in astrophysics and related fields.
Eligibility
Applicants should:
- Hold a PhD in astronomy, physics, machine learning, statistics, or a related field
- Have completed their PhD within the past ~7 years (flexible)
- Demonstrate interest in developing or applying AI/ML methods in scientific research
Researchers with backgrounds in astrophysics, machine learning, statistics, applied mathematics, or computational science are encouraged to apply.
Questions? Email us at: kaai-fellows@andrew.cmu.edu