Senior Staff Engineer, AI Safety & Agent Security (Remote)

Cohere; United States; Remote; USD 170,000 - 250,000; Full time

Job summary

Cohere is hiring for a Member of Technical Staff in the Safety for Agents team. The role focuses on data generation, post-training algorithms, and evaluation methods to ensure safer, more trustworthy LLMs.

You will collaborate with cross-functional ML, product, and policy teams and work on scalable data workflows in a remote-friendly setup with ~50% overlap with the UK/EU timezone. Ideal candidates bring strong ML experiments experience, distributed training know-how, and excellent Python and ML

Qualifications

• Strong statistical skills and experience evaluating experiments for data collection and model performance.
• Excellent software engineering skills with robust debugging and tooling experience.
• Experience designing and conducting data collection tasks, including annotator workflows.
• Hands-on experience training large language models on distributed infrastructures.
• Proficiency in Python and ML frameworks (PyTorch, TensorFlow, JAX).

Responsibilities

• Develop data generation and evaluation pipelines to improve Safety for Agents capabilities.
• Design experiments to measure model safety, bias, and generalization.
• Collaborate with ML, product, and policy teams to align safety objectives.
• Coordinate with data annotation teams to ensure high-quality labeled data.

Skills

Statistical analysis

Software engineering

Data collection design

LLM training

Python

PyTorch / TensorFlow / JAX

Tools

Distributed training infra

Job description

Cohere is hiring for a Member of Technical Staff in the Safety for Agents team. The role focuses on data generation, post-training algorithms, and evaluation methods to ensure safer, more trustworthy LLMs.

You will collaborate with cross-functional ML, product, and policy teams and work on scalable data workflows in a remote-friendly setup with ~50% overlap with the UK/EU timezone. Ideal candidates bring strong ML experiments experience, distributed training know-how, and excellent Python and ML