Veldra builds verification and observability infrastructure for production pool operators. We enable SV2 adoption by providing fail-closed safety, deterministic policy evaluation, and structured reason codes for every decision.
Eliminate template ambiguity in SV2 mining pools. Every rejection must be self-explanatory with a stable reason_code and measured context. No opaque failures.
Production-grade Rust (tokio, serde, tracing, clap). Deterministic evaluation. Structured observability from day one. Explicit errors with canonical reason codes. Stable machine-readable interfaces.
SV2 miner-facing gateway with template verification, fail-closed prevhash switching, and a two-event share lifecycle.
Noise NX handshake, SV2 frame decode, channel management, share validation, job distribution.
Template evaluation against policy.toml, fee tier selection, consensus safety checks.
Upstream template polling, deduplication, TemplatePropose packaging.
Shared types (GatewayReason, ShareAccepted, ShareForwardResult), NDJSON wire format.
Same input, same policy, same verdict. Always.
50ms verdict timeout, 5s stale hold, explicit disconnect.
Every decision logged with tracing spans, metrics, events.
Every disconnect, every rejection has a machine-readable reason_code.
Reproducible, scalable, auditable. No hidden state accumulation.
Designed for the SV2 standard job server profile, not centralized control.
jarrondeng@veldra.org