Architect, Language Model Pretraining

Bison; San Francisco (CA); On-site; USD 190,000 - 240,000; Full time

Benefits offered by this job

Comprehensive medical, dental, vision,

FSA plans

Competitive compensation and 401(k)

Relocation and immigration support on

In-office snacks and meals provided

Unlimited PTO and company holidays

In-person team in San Francisco with a

Job summary

Zyphra, an AI company based in San Francisco, seeks a Research Engineer - Language Model Pre-Training to help shape the language model roadmap through end-to-end pretraining. You will work closely with our pretraining team to translate insights into next-generation models.

Responsibilities include running large-scale training, optimizing the pretraining stack, collecting and evaluating datasets, and conducting architecture and methodology research including optimizer ablations.

Qualifications

• Strong ML problem-solving and training model expertise.
• Experience with training on large-scale multi-node GPU clusters.
• Understanding of training pipelines including model/data parallelism and distributed optimizers.
• Experience with rigorous ablations and experiments.
• Experience with large-scale data processing pipelines.
• High proficiency with Python.
• Ability to dive into large codebases quickly.
• Published ML research in well-known venues is a plus.
• Postgraduate degree in CS/EE/Math/Physics.

Responsibilities

• Large-scale training runs and model parallelization.
• Performance optimization of our pretraining stack.
• Dataset collection, processing, and evaluation.
• Architecture and methodology research, including optimizer ablations.

Skills

Large-scale training

PyTorch

Python

Distributed optimization

Codebase navigation

Education

Postgraduate degree in CS/EE/Math/Physics

Tools

PyTorch

Job description

Zyphra, an AI company based in San Francisco, seeks a Research Engineer - Language Model Pre-Training to help shape the language model roadmap through end-to-end pretraining. You will work closely with our pretraining team to translate insights into next-generation models.

Responsibilities include running large-scale training, optimizing the pretraining stack, collecting and evaluating datasets, and conducting architecture and methodology research including optimizer ablations.