poolside

Member of Engineering (Pre-training / Data Engineering)

poolside
Apply
7 months ago
Remote, EMEA or London, United KingdomSenior

Responsibilities

  • Build and maintain high-performance data pipelines processing trillions of tokens.
  • Deliver diverse, high-quality datasets for pretraining foundation models and coding agents.
  • Engineer ingestion, deduplication, streaming, data modeling, algorithmic sorting, and distributed pipeline optimization at petabyte scale.
  • Collaborate with Pretraining, Posttraining, Evals, and Product teams to align dataset quality with model capabilities and downstream use cases.

Requirements

  • Strong experience building production-grade distributed data systems for machine learning.
  • Experience with orchestration tools such as Slurm, Airflow, or Dagster.
  • Experience with observability and reliability tools such as Grafana and Prometheus.
  • Experience with Git, Docker, Kubernetes, and cloud-managed services.
  • Experience with batch inference, such as vLLM, and large-scale GPU clusters or distributed pipelines.
  • Expert-level Python knowledge, strong algorithmic foundations, and the ability to write clean, maintainable code.
  • Proficiency with Polars, Dask, or PySpark.
  • Preferred experience includes building trillion-scale state-of-the-art pretraining datasets, translating research to production at scale, OCR, web crawling, evaluations, or pretraining large language models.

Benefits

  • Fully remote work with flexible hours.
  • 37 days per year of vacation and holidays.
  • Health insurance allowance for the employee and dependents.
  • Company-provided equipment.
  • Well-being, continuous-learning, and home-office allowances.
  • Frequent team gatherings, including three days of in-person collaboration in Paris each month and annual off-sites.
  • Diverse and inclusive people-first culture.

Tech Stack

Apache AirflowDockerGitGrafanaKubernetesPrometheusPython

Categories

Data EngineeringML Engineering
poolside

About poolside

201-500 employees
Contact me