Internship on Fault-tolerant Scheduling Strategies for Iterative Algorithms

About the role

This 6‑month internship focuses on designing fault‑tolerant scheduling strategies for iterative algorithms running on large‑scale high‑performance computing platforms. The work will be carried out within the ROMA team at the LIP laboratory of ENS Lyon and may lead to a PhD thesis starting in Fall 2025.

Key responsibilities

• Investigate the impact of fail‑stop and silent errors on iterative solvers.
• Develop and evaluate checkpoint‑restart policies and algorithm‑based fault‑tolerance (ABFT) techniques.
• Design scheduling strategies that balance performance and resilience.
• Implement prototypes and conduct experiments on HPC clusters.
• Document results and prepare a one‑page research summary.

Required profile

• Master’s (M2) student in Computer Science, Applied Mathematics or a related field.
• Strong interest in high‑performance computing and fault tolerance.
• Ability to work independently and collaborate with researchers.
• Good written and oral communication skills in English.

Required skills

• Knowledge of iterative algorithms and numerical methods.
• Familiarity with fault‑tolerant concepts such as checkpoint‑restart and ABFT.
• Experience with HPC environments (clusters, parallel programming).

What we offer

• Mentorship from leading researchers (Anne Benoit, Yves Robert, Emmanuel Agullo).
• Potential continuation into a funded PhD position (NumPEx PEPR program).
• Access to state‑of‑the‑art computing facilities at ENS Lyon.