How AI has shifted the bottleneck in the development cycle
With LLMs, implementation is no longer where agile cycles lose time: the cost has moved to planning and verification, and living documentation helps with both.
Hi, I'm Javi.
I'm a data engineer who builds AI agents and the data platforms that feed them — turning long-running engineering work into small, checkpointed loops.
Selected work
-
Gubia
A meta-harness that turns a goal into an auditable plan and drains it task by task through CLI coding agents, reviewing the evidence at each checkpoint.
-
Data Pipeline Platform
A batch data platform that ingests raw sources and serves analytics-ready, versioned datasets to teams and AI agents.