Software Engineer, Machine Learning Infra in San Francisco, CA

$80K - $100K(Ladders Estimates)

Pinterest   •  

San Francisco, CA

Industry: Advertising & Marketing Services

  •  

Not Specified years

Posted 878 days ago

This job is no longer available.

Numerous teams at Pinterest use machine learning in production, but many of them currently have custom infrastructure. We're building a unified system to serve and train production models, providing a common ML platform for the company. You'll be a key engineer in designing and developing the training pipeline for this system, and you'll later build upon this platform to accelerate all ML development at Pinterest.

What you’ll do :

  • Lead design and development of an ML training system that will be used for ad relevance prediction, pin scoring, and feed ranking
  • Work with multiple teams to understand their requirements and devise a solution that they can satisfy all their use cases
  • Develop diagnostic tools, monitoring systems, and evaluation methodologies that can be shared across our ML use cases

What we’re looking for :

  • Experience addressing the challenges of production machine learning systems, such as managing/verifying upstream data dependencies, enabling experimental development, automating model training, and testing models
  • Experience building offline data pipelines, with a solid understanding of how Hadoop MapReduce works and practical experience with higher-level tools like Hive, Cascading, or Scalding
  • An understanding of Web-scale systems and ML serving infrastructure

Valid Through: 2017-6-19