Summary

This role combines data analysis and engineering to improve Perplexity’s web crawling, indexing, and search-and-answer systems. You’ll build evaluation datasets and pipelines, train small production models, and run experiments to improve quality.

Responsibilities

• Find and diagnose crawling pipeline quality issues
• Train small models to improve pipeline quality
• Build training datasets and llm-as-a-judge labeling pipelines
• Improve page selection algorithms for indexing
• Design and analyze experiments validating improvements