Cohere: Member of Technical Staff, Post-Training

Cohere is seeking a Member of Technical Staff, Post-Training to join its team in New York, NY on a full-time basis. In this role, you will advance the state of the art for foundation model post-training and ship frontier models directly to production, bridging deep learning research with scalable enterprise software.

Cohere operates with one of the highest compute-to-engineer ratios in the AI industry, maintaining an environment without rigid boundaries between research and production engineering. Rather than isolating experimental ideas, every technical team member writes production-grade code, tests concepts across supercomputing clusters, and deploys high-impact models for enterprise adoption. While aligned with the New York City office, the position supports remote-friendly flexibility with no mandatory in-office attendance requirement.

• Scalable Software Architecture: Design and implement high-performance, fault-tolerant software for training large-scale models.
• Post-Training Optimization: Consistently post-train models to achieve state-of-the-art (SOTA) benchmarks on real-world business tasks.
• Algorithmic Development: Innovate and implement techniques to enhance training cycle results across Supervised Fine-Tuning (SFT) and Reinforcement Learning (RL) regimes.
• Cross-Functional Collaboration: Partner closely with dedicated specialist teams, including Code and Agentic systems, to produce well-rounded, capable models.
• Supercompute Experimentation: Research, implement, and validate novel ideas across dedicated data infrastructure and distributed accelerators.

• Core Engineering: Exceptionally strong software engineering skills with an emphasis on production quality.
• ML Frameworks: High proficiency in Python and modern frameworks, including PyTorch, JAX, and XLA/MLIR.
• Distributed Infrastructure: Hands-on experience with orchestration and distributed training frameworks, such as Kubernetes, Slurm, and Ray.
• At-Scale Training: Demonstrated track record executing large-scale distributed training strategies.
• Post-Training Specialization: Direct experience managing the post-training phase of large models, with a dedicated focus on performance optimization.
• Research Record (Bonus): Publications at premier machine learning venues such as NeurIPS, ICML, ICLR, MLSys, ACL, or EMNLP.

• Health and Wellness: Comprehensive health and dental coverage, including a dedicated mental health budget.
• Retirement Programs: 401(k), RRSP matching, or pension scheme participation based on location.
• Parental Support: 100% parental leave top-up for up to six months for either parent.
• Meal Allowance: A weekly lunch stipend of $75/£75 (or local currency equivalent).
• Workplace Flexibility: Remote-friendly work options alongside access to key global offices, with zero minimum in-office requirements.

• Highlight Distributed Scaling: Explicitly detail cluster configurations you have managed, including accelerator types, interconnect considerations, and frameworks like Ray or Slurm.
• Showcase Alignment and Optimization: Detail specific SFT or RL implementations where your architectural adjustments improved training efficiency or downstream model metrics.
• Reference Public Technical Artifacts: Highlight any open-source model repositories, systems code, or peer-reviewed publications relevant to large-scale deep learning.

This post-training position offers a high-impact opportunity to push the limits of generative AI models backed by extensive supercomputing infrastructure and a collaborative research environment.