From 35 Million Rows of Data to a Datalake: What One Person Can Build with Agentic AI

​Most teams can source data far faster than they can interpret it and that gap is where analytics projects stall.

​In this session, we'll walk through River Signal, my personal passion project, and a working watershed-intelligence platform built with Claude Code: 35M+ records ingested from 43 public data sources, then transformed, enriched, and landed in a PostgreSQL/PostGIS lake. It follows a medallion architecture surfacing insights through a map-first React analytics UI backed by an LLM reasoning layer.

​I'll walk through real prompts, architecture decisions, and the iteration loop that took it from raw public APIs to a product that answers "what do these observations imply for watershed management?" instead of just returning rows.

​If you're an AI strategy leader deciding how agentic coding changes your data platform roadmap or an engineer eager to learn more about what's possible with Claude Code, this is a concrete, end-to-end look at what a individual or small team can now ship.