Member of Technical Staff, Integration/RL Team (Research Engineer)

Who are we?

Cohere is the leading security-first enterprise AI company. We build cutting-edge foundation AI models and end-to-end products that are designed to solve real-world business problems.

We’re training and deploying frontier models for enterprises who are building AI systems. We believe that our work is instrumental to the widespread adoption of AI and we are looking for folks that want to be part of that.

We obsess over what we build. Each one of us is responsible for contributing to increasing the capabilities of our models and the value they drive for our customers. Cohere is a team of researchers, engineers, designers, and more, who are all passionate about their craft.

We are a global technology company headquartered in Toronto with key offices in London, New York City, San Francisco, Montreal, Paris, Berlin and Seoul. Join us!

Role Overview:

The integration team is responsible for developing and scaling machine learning algorithms and infrastructure for LLM post-training, with a focus on large-scale, distributed RL methods. We strive for excellence in both engineering and science by meticulously designing experiments and design docs. While tasks are assigned according to everyone’s expertise, there is a global team effort to write production code and support the team research efforts, depending on individual interests and organizational needs.

In particular, this role aims to enhance the global quality of the post-training codebase by implementing new tools to ease and support research, optimizing post-training algorithms, and scaling distributed RL to unprecedented levels.

Please Note: We have offices in London, Paris, Toronto, San Francisco, New York but we are also remote-friendly! Applicants for this role may work anywhere between UTC−06:00 and UTC+01:00.

Key Responsibilities:

Design and write high-performing and scalable software for training models.

Develop new tools to support and accelerate research and LLM training.

Coordinate with other engineering teams (Infrastructure, Efficiency, Serving) and the scientific teams (Agent, Multimodal, Multilingual, etc.) to create a strong and integrated post-training ecosystem.

Craft and implement techniques to improve performance and speed up our training cycles, both on SFT, offline preference, and the RL regime.

Research, implement, and experiment with ideas on our cluster and data infrastructure.

Collaborate, Collaborate, and Collaborate with other scientists, engineers, and teams!

Qualifications:

Extremely strong software engineering skills.

Value test-driven development methods, clean code, and strive to reduce technical debts at all levels.

Proficiency in Python and related ML frameworks such as JAX, Pytorch and/or XLA/MLIR.

Experience using and debugging large-scale distributed training strategies (memory/speed profiling).

[Bonus] Experience with distributed training infrastructures (Kubernetes) and associated frameworks (Ray).

[Bonus] Hands-on experience with the post-training phase of model training, with a strong emphasis on scalability and performance.

[Bonus] Experience in ML, LLM and RL academic research.