🛠 Tech Stack

💼 About This Role

You'll advance the state of the art for model post-training and ship models to production. You'll bridge the gap between research and production with one of the highest compute-to-engineer ratios. You'll design high-performant scalable software and implement cutting-edge training techniques.

🎯 What You'll Do

• Design and write high-performant scalable software for training models.
• Post-train models to reach SOTA level performance.
• Coordinate with specialist teams for strong all-encompassing performance.
• Craft and implement techniques for SFT and RL training cycles.
• Research and experiment on supercompute and data infrastructure.

📋 Requirements

• Extremely strong software engineering skills.
• Proficiency in Python and ML frameworks like JAX, PyTorch.
• Experience with distributed training infrastructures (Kubernetes, Slurm, Ray).
• Hands-on experience training large models at scale.

✨ Nice to Have

• Publication at top-tier venues (NeurIPS, ICML, ICLR, etc.).
• Experience with post-training phase and performance optimization.

🎁 Benefits & Perks

• 🤝 Open and inclusive culture
• 🧑‍💻 Cutting-edge AI research team
• 🍽 Weekly lunch stipend and in-office meals
• 🦷 Full health and dental benefits with mental health budget
• ✈️ 6 weeks vacation (30 working days)