Software Engineering Daily Context-Aware SQL and Metadata with Shinji Kim
59 snips
Sep 4, 2025 
Shinji Kim, the founder and CEO of Select Star, discusses the challenges of metadata curation and the innovative solutions his company offers. He elaborates on the importance of maintaining context in data-rich environments and how Select Star creates a constantly updated knowledge graph. Shinji also touches on the role of large language models in generating accurate SQL queries and the significance of tracking data usage metrics to enhance decision-making. His insights reveal emerging trends in metadata management that could transform how organizations access and utilize their data.
AI Snips
Chapters
Transcript
Episode notes
Leverage Popularity To Cut Waste
- Use popularity and lineage together to find trusted datasets and reduce cloud costs.
- Drop or archive unused models and update BI artifacts informed by these metrics.
Hydrate AI With Curated Metadata
- Feed popularity scores and example queries into AI agents to improve natural language to SQL accuracy.
- Let agents use curated semantically defined models instead of relying solely on raw schema.
POC Shortcuts Don't Scale For Enterprise AI
- POCs often hardcode schema, synonyms, and examples to make AI work, but that approach doesn't scale.
- Continuous metadata platforms are required to maintain context for enterprise AI at scale.
