Latest AI JobsBrowse Jobs
Cohere logo

Member of Technical Staff, Training Performance Engineer

CohereLondonModeling
Apply on company site
Salary
$250K - $535K CAD
Job type
Full-time
Work mode
Remote
Posted
20 Feb 2025
Status
Open · checked 7d ago

About this role

The Member of Technical Staff, Training Performance Engineer position at Cohere is a full-time role based in London with remote flexibility, perfect for software professionals passionate about optimizing large-scale AI infrastructure. Candidates should reside in locations aligned with Eastern Time to Central European Time hours for smooth collaboration.

Role Overview

As a Performance Engineer in the Pre-Training team, the individual handles the optimization of sophisticated language models and underlying systems. The core focus centers on boosting model training metrics, including training throughput and accelerator utilization. The position bridges software engineering, machine learning, and low-level kernel development to build reliable systems and elevate model capabilities. Responsibilities include spotting and clearing performance roadblocks, building cutting-edge training and profiling tools, and advancing natural language processing capabilities.

Key Responsibilities

Design and write high-performant, scalable software dedicated to model training.
Comprehend architectural choices and judge their direct impact on training quality and throughput.
Write low-level CUDA and Triton kernels to extract maximum performance from accelerators.
Research, test, and implement experimental concepts utilizing supercompute and data infrastructure.
Learn alongside leading researchers in the artificial intelligence sector.

Requirements

Exceptionally strong software engineering abilities.
Proficiency in Python alongside machine learning frameworks like PyTorch, JAX, and XLA/MLIR.
Hands-on experience writing GPU kernels utilizing Triton, CUDA, or similar technologies.
Practical background deploying large-scale distributed training strategies.
Familiarity with autoregressive sequence models, specifically Transformers.

Nice to Have

Published papers at top-tier venues such as NeurIPS, ICML, ICLR, AIStats, MLSys, JMLR, AAAI, Nature, COLING, ACL, or EMNLP.

Skills and Tools

Python
PyTorch
JAX
XLA/MLIR
CUDA
Triton
Distributed Training
Transformers

Salary and Benefits

Annual salary range between 250,000 and 535,000 CAD.
Weekly lunch stipend of $75/£75 or local currency equivalent.
Comprehensive health and dental benefits, featuring dedicated mental health funding.
RRSP matching, 401K, and Pension Scheme options.
Full parental leave top-up covering up to 6 months for either parent.
Annual enrichment benefits for fitness, arts, culture, and workspace improvements, plus an education and learning stipend.
Six weeks of paid vacation, equating to 30 working days.
Travel budget for visiting other company offices and attending annual company offsites.
Home office setup stipend of $500 for remote workers, or co-working and daily lunch benefits for office-based staff.

About the Company

Cohere operates as a leading security-first enterprise AI company building advanced foundation AI models and end-to-end products to address real-world business challenges. Headquartered in Toronto with global offices spanning London, New York City, San Francisco, Montreal, Paris, Berlin, and Seoul, the organization focuses heavily on driving wide enterprise AI adoption.

How to Apply

Interested candidates should submit their application directly on the official Cohere careers page.

Skills and tech stack

Questions candidates ask

Is this position remote?

Yes, the role offers remote flexibility, though candidates should align with Eastern Time to Central European Time working hours.

What programming languages and frameworks are required?

Applicants need strong software engineering skills with proficiency in Python and machine learning frameworks such as PyTorch, JAX, and XLA/MLIR.

What kind of benefits does Cohere provide for full-time employees?

Perks include comprehensive health and dental coverage, mental health funding, a pension scheme, 401K or RRSP matching, 6 weeks of paid vacation, parental leave top-up, and stipends for home offices, learning, and weekly lunches.

Where is the company located?

Cohere is headquartered in Toronto with major offices in London, New York City, San Francisco, Montreal, Paris, Berlin, and Seoul.

How should I apply for this job?

Applications must be submitted through the official Cohere careers page.

Ready to apply?

The application is on the official Cohere careers page.

Apply on company site