SC 012 — "We're So Back" — The Analytics Engineering Angle

Summary

After a year of community stagnation, identifies five emerging capabilities signaling analytics engineering’s resurgence:

1. The Ultimate Rubber Duck — GitHub Copilot Chat as LLM-enabled development, maintaining workflow continuity without context-switching
2. Modular Analytics — using synthetic datasets (Faker, Factory Boy) to make analytics work portable, shareable, and portfolio-worthy
3. dbt Unchained — Josh Wills’ dbt-duckdb adapter extending dbt beyond transformation into ingest and serve, reducing vendor dependencies
4. BI-as-Code — Evidence.dev combining Markdown with SQL-powered charts; version control, deployment environments, and LLM integration
5. Paved Path to Prod — MotherDuck streamlining production deployment via connection string changes (main.duckdb to md:main.duckdb)

Key Arguments

• Analytics engineering hit a wall but breakthroughs are emerging simultaneously
• Modular analytics would democratize knowledge sharing and enable portfolio-building — currently, work stays locked inside companies
• dbt-duckdb breaks dbt out of its transformation-only box
• BI-as-code gains version control, environments, and LLM advantages for free
• The production deployment gap is closing — MotherDuck makes it a config change

Writing Style Notes

Optimistic and energized — “We’re So Back” captures the vibe perfectly. Each capability gets a memorable label. The enthusiasm is genuine, grounded in specific tools rather than vague trends. Recommends Fundamentals of Data Engineering for its lifecycle-based framework.

Connections

• index — part of the Sanity Check body of work
• index — father’s Squarely puzzle book mentioned as newly published
• 2026 04 03 headless bi — BI-as-code connects to headless BI movement