Zyphra

Research Engineer - Language Model Pre-Training

3mo ago
USASenior
Zyphra

Research Engineer - Language Model Pre-Training

3mo ago
USASeniorpytorchpythondistributed systemsmachine learningdata processingmodel training

Research Engineer role focusing on language model pre-training and optimization at an AI company.

Responsibilities

  • As a Research Engineer - Language Model Pre-Training , you'll shape our language model roadmap through end-to-end pretraining development. You will work extremely closely with our pretraining team, who will integrate your insights into our next-generation models.

Requirements

  • Strong engineering aptitude for rapidly implementing reliable and robust systems
  • Can rapidly learn new fields and are excited to implement new ideas
  • Excellent communication and collaboration skills, and can work effectively on both research and engineering implementation at scale
  • Deep expertise and intuition for solving machine learning problems and training models
  • Experience with training on large-scale (multi-node) GPU clusters
  • Deep understanding of model training pipelines – including model/data parallelism, distributed optimizers, etc.
  • Strong grasp of proper experimental methodology for running rigorous ablations and other hypothesis testing
  • Understanding of large-scale, highly parallel data processing pipelines
  • High proficiency with PyTorch and Python.
  • Strong ability to dive into large pre-existing codebases and rapidly get up to speed
  • Published machine learning research in well-respected venues is a plus
  • Postgraduate degree in a scientific subject (Computer Science, EE/EECS, Math, Physics)

Conditions

  • Comprehensive medical, dental, vision, and FSA plans
  • Competitive compensation and 401(k) plan
  • Relocation and immigration support on a case-by-case basis
  • In-office snacks and meals provided
  • Unlimited PTO and company holidays
  • In-person team in San Francisco with a collaborative, high-energy environment

Other

  • 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
  • Our research methodology is to make grounded, methodical steps toward ambitious goals. Both deep research and engineering excellence are equally valued
  • We strongly value new and crazy ideas and are very willing to bet big on new ideas
  • We move as quickly as we can; we aim to minimize the bar to impact as low as possible
  • We all enjoy what we do and love discussing AI