
OpenAI engineer Bonnie Xu built Kepler, an internal AI data analyst agent that queries over 600 petabytes of data. The system overcomes context window limits using MCP, automated code crawling, and RAG. Kepler also uses scoped semantic memory for self-learning and AST-based LLM grading for a regression-free evaluation pipeline.
Tap to vote and see what everyone thinks.
Summary by ByteBrief
Cursor · Origin