Repo

GenDB-V

Builds verified Rust query engines for a schema or hot query template by translating SQL to a fixed spec, then generating code and proofs checked by Verus.

GitHub ↗Nov 1, 2026Paper ↗

  • llm
  • formal-verification
  • sql
  • query-optimization
Pipeline diagram showing GenDB-V synthesis, verification, and trust boundaries
Pipeline diagram showing GenDB-V synthesis, verification, and trust boundaries

What it is

Synthesizes workload-specific Rust database engines from SQL and formally verifies them with Verus. Reported speedups reach 61.3× over DuckDB and 28.3× over Umbra. Generate a workload- and hardware-specific database engine from SQL, then prove the produced Rust crate returns correct answers and is memory-, arithmetic-, and race-safe.

Key results

  • Up to 61.3× over DuckDB and 28.3× over Umbra on SEC-EDGAR
  • 51.6× over DuckDB and 22.0× over Umbra on TPC-H 22 queries
  • Verus proves correctness, memory safety, arithmetic safety, no data races

Shows AI can synthesize database engines that are both formally verified and substantially faster than mature systems on targeted workloads.

One useful ML email a week.

Worth your time, not all of your time.

We track new ML work across papers, code, and data, then keep only the pieces worth a closer look. Selected, not collected.

Papers.Data.Code

The Weekly ML Signal

Selected, not collected.

One useful ML email a week: the papers, repos and datasets worth your attention.

Worth your time, not all of your time.