Built by engineers who've lost a Sunday to a renamed column.
In early 2023, Michael Adeyemi was the data platform lead at a logistics analytics company in Miami. The company pulled from seventeen vendor APIs. Every two or three months, one of those vendors would push a schema change with no migration notice — a renamed field, a dropped column, a type widened from INT32 to BIGINT. The pipeline would keep running. The data would keep landing. The dashboards would go silently wrong.
The specific incident that broke it open: a freight-rate vendor renamed order_amount to total_value on a Friday evening. The revenue dashboard zeroed out over the weekend. Nobody caught it until Monday standup. Three days of stale data, one very bad management review.
Michael spent the next two months building a schema-inspection layer on top of their existing orchestrator. It worked. He left in late 2023 to turn it into Queryvine. Priya Nair joined from her streaming infrastructure work at a fintech — she'd spent years dealing with Avro schema evolution in Kafka topics and had a clear picture of what the detection layer needed to look like. Marcus Okafor came on to drive the connector ecosystem and keep the docs honest.
We're bootstrapped, three people, based in Miami. We work in the open — our SDK is on GitHub, our changelog is public, and our docs are the primary product surface.
How we work
Config over clicks
Pipelines are YAML files in your repo. They go through PR review, run through CI, and deploy through the same tooling as everything else in your infrastructure. We don't offer a drag-and-drop builder and don't intend to.
Transparency by default
Every drift event, every reroute decision, every schema version stored. The full lineage of what changed and when is inspectable — via CLI, API, or the schema history endpoint. Nothing happens silently.
Fail loudly, not quietly
When an upstream schema change has no matching drift rule, Queryvine pauses the pipeline and sends an alert. A paused pipeline is recoverable. Three days of silently corrupt data in your warehouse is not.
Scope honesty
Queryvine is not an ETL tool. It does not extract or load data on its own. It orchestrates pipelines that already move data and adds schema-drift awareness to every hop. Knowing what we're not is as important as knowing what we are.
The team
Ten years in data engineering. Previously architected streaming pipelines at a fintech processing millions of transactions daily, where Avro schema evolution in Kafka topics was a daily operational concern. She designed Queryvine's fingerprinting engine and drift rule evaluation system.
Contributor to several open-source orchestration projects including Apache Airflow provider libraries. Specializes in schema evolution strategies, distributed data systems, and the operational patterns that keep large-scale pipelines healthy without constant manual intervention.
We'd love to hear how your pipelines break.
Seriously — schema drift war stories help us build a better product. Email us, open a GitHub issue, or just start a free trial and see what we catch.