// the find
google/fhir
FHIR Protocol Buffers
Google's implementation of the FHIR healthcare data standard using Protocol Buffers as the underlying representation. It gives you type-safe FHIR R4/R5/STU3 resources in C++ and Go, with JSON parsing/printing, FHIRPath validation, and support for custom Implementation Guides. This is for teams building healthcare data pipelines at scale who need something more structured than raw JSON and are already in the Google ecosystem.
Proto encoding means significantly smaller payloads on disk and in transit compared to FHIR JSON — relevant when you're processing millions of patient records. FHIRPath is fully implemented with an ANTLR grammar and a proper expression evaluator, not a regex hack. The profile/Implementation Guide toolchain lets you generate protos from custom IG definitions, which is the right architecture for constrained datasets. Active R5 support was added recently and the last push was yesterday, so this isn't abandoned.
Bazel-only build system — if your org isn't already on Bazel you're in for a painful onboarding, and there's no cmake or Makefile fallback for the C++ layer. The Go bindings are auto-generated and thin; don't expect idiomatic Go. The README punts almost entirely to a separate examples repo, so discoverability of actual API surface is poor. DSTU2 support exists in the Go protos but isn't in the C++ layer, which will bite anyone working with legacy EHR data still on DSTU2.