Software Engineer, Data Acquisition

About the role: We are looking for a Software Engineer to lead the "Tokens: Data Acquisition" team, which owns the problem of acquiring all of the available data on the internet through a large scale web crawler, and through data partnerships. Most of Anthropic's research and product builds on top of the best pretrained models that we can produce, which in turn rely on having the best pretraining data. Successfully scaling our data corpus is critical to our continued efforts at producing the best pretrained models.

The team's responsibilities are as follows:

Develop and maintain an internet scale web crawler responsible for crawling for accessible internet data Build required pipelines to quickly ingest data from potential partners for data quality assessments or from other sources Build a set of 1-off deep crawlers to improve recall for data from especially beneficial sources Responsibilities: Develop and maintain our large-scale web crawler Build pipelines for data ingestion, analysis, and quality improvement Build specialized crawlers for high-value data sources Build tools for improving the observability and debuggability of crawler system Collaborate with team members on improving data acquisition processes Participate in code reviews and debugging sessions You may be a good fit if you: Believe that advanced AI systems could have a transformative effect on the world, and are interested in helping make sure that transformation goes well Are interested in building a large scale system to acquire all openly accessible information on the internet Strong candidates may also: Extensive experience with building and running large distributed systems Familiarity with the non-technical tradeoffs of internet-scale crawling (data privacy, robots.txt adherence, etc.) Technical expertise: Quickly understanding systems design tradeoffs, keeping track of rapidly evolving software systems Familiarity with Cloud-based compute and storage solutions Familiarity with Python


#J-18808-Ljbffr