The person behind all of it
Rusty Conover
Founder & CEO
Rusty has spent more than two decades building data infrastructure that has to stay up and stay fast. He was Head of Cloud Data Platform at Schonfeld Strategic Advisors, and held senior engineering and CTO roles at J.P. Morgan and Two Sigma, running petabyte-scale analytics for quantitative trading. He’s also a repeat founder — he co-founded and was CTO of GearBuyer / InfoGears, scaling it past $21M in annual transactions.
Query.Farm grew out of that work. DuckDB is a superb engine, but production teams kept needing the same things around it: extensions, connectors, and a distribution hardened for real workloads. So he started building them — shellfs in May 2024, and 31 more since. Today he contributes directly to DuckDB’s core, publishes the extensions this site documents, and builds the platform layer on top of them.
Query.Farm is founder-led and guided by a board of advisors.
Three layers, one stack
Everything Query.Farm makes builds on DuckDB — and on each other.
Extensions
32 DuckDB extensions — connectors, hashing, geospatial indexing, scripting, and more.
Learn more →Haybarn
An independent, hardened distribution of DuckDB — signed builds, clients, and a browser engine.
Learn more →VGI
The Vector Gateway Interface: expose any external system as a queryable DuckDB table, in any language.
Learn more →What we stand for
Open source, and honest about what isn’t
Most of what we make is open source, including the extensions this site documents. We also build commercial and closed-source products — and we tell you which is which.
We contribute to the core, not just the edges
Our work lands in DuckDB itself — Arrow integration, the C API, scalar functions, statistics, serialization — so the tools we build on top stand on solid ground.
Built to be run in production
Signed builds, verifiable artifacts, real load counts, and a hardened distribution. Not demos — software you can ship.
Useful before it’s a platform
We ship the part that does the job and skip the platform detour. You shouldn’t have to stand up infrastructure to answer one question.
From one extension to a stack
The first extension
shellfs ships — pipe shell commands straight into DuckDB SQL, and the steady cadence begins.
Crypto, geometry, and scripting
Within weeks: cryptographic hashing, space-filling curves for spatial ordering, Arrow Flight connectivity, and an embedded scripting language.
Reaching other systems
HTTP serving and client calls, LLM prompting from SQL, and Apache DataSketches — which alone now sees millions of loads.
Scale arrives
A full JavaScript engine, probabilistic filters, and fast non-cryptographic hashing push loads into the millions every week.
Past thirty extensions
Pentagonal geospatial indexing, an ADBC scanner, JSONata queries, templating, and string tooling round out the catalog.
VGI launches
The Vector Gateway Interface turns any external system into a queryable table, with SDKs across Go, Java, Python, Rust, TypeScript, and Kotlin.
Geometry gets smaller
geosilo lands a first-class column type that delta-encodes coordinates as scaled integers and pairs with ZSTD — about 21% smaller geometry storage.
Haybarn
An independent derived distribution of DuckDB — build-forks of Iceberg, DuckLake, Delta, and HTTPFS, plus clients for Python, Node, JDBC, Rust, and WebAssembly.
Orchard and Cupola
Orchard opens a marketplace of VGI connectors you attach with one line of SQL, and Cupola puts a DuckDB analyst workbench — SQL, pivots, charts — in a browser tab.
Chips show each extension’s loads over the last 90 days. Explore them all on the extensions page.
Building on DuckDB? Let’s talk.
Whether you want to adopt our extensions, run Haybarn in production, or need a hand with a hard data problem — we’d love to hear from you.
Two ways in
Book a call, or start with the catalog and see whether something we already ship does the job.