About us

SynnoDB

We synthesize database engines built for one workload: for OLAP and for ETL. Orders of magnitude faster and cheaper.

Where we are

The engines work.
Let's build it together.

Every analytical workload today runs on an engine built to be ready for any query anyone might write, and you pay for that generality on every run. SynnoDB synthesizes an engine that only does your job, and it now covers both halves of the stack:

  • OLAP. Engines that run TPC-H 21.78× faster than DuckDB / Umbra / ... for about ~$80 of synthesis cost. Run one in your browser.
  • ETL. Engines synthesized straight from a dbt project: 12.2× faster on a 16 GiB machine, and 990 GiB, 19.5 billion rows, pushed through the same box DuckDB cannot finish 82 GiB on. See the numbers.

Published research, running code, reproducible benchmarks. What's left to build is the company around it - self-serve synthesis, hosting, and the same results on workloads that aren't benchmarks.

We're looking for partners to build it with.

Design partners

Try it on your workload

A dbt project or a set of SQL queries is enough. We synthesize an engine for it, you see the numbers first, and the product gets shaped around what you actually run.

Investors

Back the spin-out

We're talking to investors who know data infrastructure and want in while the technical direction is still being set.

Engineers

Build it with us

Database internals, query compilation, and LLM-driven code synthesis - with real ownership from day one.

Research & ecosystem

Push it further

Joint research, benchmarks, and integrations across the data stack. We stay close to the research community this came out of.

Meet the Team

Built by database researchers.

Authors of the Bespoke OLAP research.

Tell us about your workload

A few SQL templates + your schema is enough for us to estimate what SynnoDB could do for you.

Send us your workload →

Authors of the Bespoke OLAP research