Member of Technical Staff, Post-Training, RL

About the Company

Mirendil is a tech-first company focused on solving core bottlenecks that unlock step-change acceleration across science and technology. We are building a frontier AI research company and training our own models end-to-end, with a team including researchers and engineers from Anthropic, Google DeepMind, xAI, OpenAI, Microsoft, Apple, and MIT.

Responsibilities

• Design and run experiments to turn base models into reasoning models by choosing training objectives and shaping data mixtures
• Develop and iterate on RL, SFT, and distillation recipes
• Scale reinforcement learning to larger models, longer trajectories, and larger compute budgets
• Develop methods for long-horizon reasoning and feedback assignment
• Work on off-policy and asynchronous training regimes
• Build robust verification pipelines to detect and reduce reward hacking
• Scale post-training recipes across different task families and domains
• Diagnose training regressions and build probes for deep empirical analysis
• Collaborate with systems, infrastructure, and data teams to move experiments to production-scale runs

Requirements

• Strong research and engineering skills
• Ability to form hypotheses about training behavior and implement them in real systems
• Experience running large-scale experiments and analyzing resulting traces
• Proficiency in designing training recipes and understanding training dynamics