💎 CRISTAL Lab

Welcome to the CRISTAL Lab at CISPA! Our mission is to develop cutting-edge AI systems that can communicate effectively, reason reliably, operate safely in complex environments, and make transparent and trustworthy decisions.

We are building a collaborative, ambitious, and inclusive research group focused on pushing the boundaries of AI alignment, reasoning, safety evaluation, and interpretability. If you are passionate about building the future of trustworthy AI, explore our open positions and application guidelines below.

🎯 Research Focus

Recent advances in foundation models offer powerful capabilities alongside core challenges in safety and reliability. CRISTAL focuses on three core, interconnected pillars:

• AI Alignment: Building foundation models and agentic systems that reliably adhere to human intent, values, and preferences.
• Safety & Robustness Evaluation: Designing systematic evaluation frameworks to anticipate failure modes, stress-test systems, and benchmark safety before deployment.
• Transparent & Interpretable Decision-Making: Developing techniques to make complex AI reasoning accountable, understandable, and continuously adaptable.

Core Research Topics: • Foundation Models • AI Alignment & Evaluation • Reasoning & Decision-Making • Agentic & Multi-Agent Systems • Continual & Lifelong Learning • Interpretability & Transparency • AI Safety & Robustness

💼 Open Positions & What We Offer

We invite applications for fully funded PhD Students, Postdoctoral Researchers, and Research Interns (BSc/MSc/PhD) in CS, ML, NLP, AI Safety, and related fields.

Why Join CRISTAL?

• Shape a Founding Lab: Help define the scientific trajectory and culture of a dynamic, new research group from day one.
• Mentorship & Autonomy: Direct 1-on-1 guidance paired with the freedom to lead independent research agendas and publish at top venues (NeurIPS, ICML, ICLR, ACL, EMNLP, COLM).
• Institutional & Compute Resources: Access to world-class computational infrastructure, travel funding for conferences, and full visa support.
• Compensation & Benefits: Full-time position with competitive salary, comprehensive health insurance, paid vacation, and pension benefits.
• Diverse Environment: International, interdisciplinary group located at CISPA in Germany, operating fully in English.

📋 Application Guidelines

Please prepare your application materials as a single PDF file (or a link to a unified folder):

1. Curriculum Vitae (CV)
2. Research Statement (2–3 pages): Detail your past research contributions, future vision, and how your interests connect to CRISTAL’s mission.
3. Representative Artifacts (Up to 3): Publications, preprints, code repositories, or technical project reports.
4. Academic Transcripts: Bachelor’s, Master’s, and PhD, as applicable.
5. References: Names and contact details of three academic advisors willing to provide reference letters upon request.
6. (Optional) Responses to Research Questions of Your Choice (1–2 pages per question, highly encouraged):
   • How would you evaluate whether a modern AI system behaves reliably and in accordance with intended goals?
   • What are important or underexplored failure modes in current state-of-the-art AI systems?
   • How would you design methods to assess or improve the transparency and interpretability of AI systems?
   • What limitations do you see in current approaches to evaluating AI safety and robustness?

Evaluation Criteria:

• PhD & Intern Applicants: Curiosity, critical thinking, technical background (projects, coursework, or early research), and growth trajectory.
• Postdoctoral Applicants: Scientific independence, publication track record in relevant venues (NeurIPS, ICML, ICLR, ACL, EMNLP, COLM, etc.), leadership potential, and alignment with CRISTAL's core vision.

📩 How to Apply

🗓️ Timeline:

• Applications are reviewed on a rolling basis. Priority review begins September 15th, 2026.
• Start dates are flexible, available as early as Fall 2026.
• For informal inquiries or pre-application questions, feel free to reach out to garbacea@uchicago.edu.