Staff Software Engineer - Platform Architecture

b.well Connected Health is addressing healthcare's fragmentation with a FHIR-based health data management platform. They are seeking a Staff Software Engineer to design and build event-driven distributed systems, own API architecture, and partner on AI integration, while ensuring the platform remains efficient and scalable.

Responsibilities:

• Design and build event-driven distributed systems on Kafka: orchestrator and worker services, schema and contract governance (AsyncAPI, CloudEvents), idempotent consumers, sagas and event choreography, and config-driven pipelines that onboard new data sources without code changes
• Own API architecture at scale across REST and a federated GraphQL gateway (WunderGraph Cosmo) that powers our SDKs, including schema composition, versioning, contract evolution, and FHIR conformance
• Partner on AI architecture and enablement: help make LLMs and agents dependable parts of the system (retrieval, evaluation, guardrails, and clear boundaries around protected health data), working alongside the dedicated Applied AI Engineering lead who owns that area
• Shape FHIR and health-data architecture: terminology services, resource modeling, clinical quality measurement (DQM and CQL), high-volume document storage, and query performance across MongoDB and ClickHouse
• Help run the technical design review process and document the decisions and standards, so significant designs are reviewed against consistent criteria and the platform stays coherent as it grows
• Stay close to the code and to production: write real software, build proofs of concept that prove out a pattern before teams adopt it, and use observability and operational data to confirm decisions hold up and to debug incidents when they don't
• Mentor senior engineers and tech leads on distributed-systems thinking, system design, and the trade-offs behind good architecture

Requirements:

• Deep expertise in large-scale distributed systems and scalable design. You've built and operated systems that handle high volume and real failure modes, and you're fluent in partitioning, back-pressure, idempotency, eventual consistency, and the trade-offs between consistency and availability
• Strong event-driven architecture experience, hands-on with Kafka: topic and partition design, schema evolution and governance, consumer groups, idempotency, and the operational side of running it in production
• Strong API design across REST and GraphQL. We run a federated GraphQL gateway, so schema composition, versioning, and contract evolution matter here as much as clean endpoint design
• A well-rounded, hands-on engineer who goes deep in at least one strongly-typed language and moves comfortably across a polyglot stack. You have strong software-design instincts and care about clean interfaces, sensible abstractions, and keeping services loosely coupled. Our backend spans Python (data and AI), TypeScript/Node.js (which runs our FHIR server), and Java/Spring Boot (FHIR processing and backend services), with Kafka, our federated GraphQL gateway, Databricks, ClickHouse, MongoDB, and AWS around it
• Sound data and storage instincts: data modeling, caching, query optimization, and how batch and streaming workloads (for example on Spark or Databricks) fit together with operational systems
• Cloud-native and security fundamentals: Kubernetes and AWS, identity and auth (OIDC, token exchange), and multi-tenant isolation in a regulated environment that handles protected health data
• You can walk someone through a system you've designed and operated end to end: how it works, where it breaks, and the trade-offs you made
• Exposure to AI as a component of a system is a plus — LLMs or agents and the practical concerns around them (retrieval, evaluation, guardrails, cost and latency). Deep AI architecture ownership sits with a dedicated Applied AI Engineering hire, so it isn't a primary requirement here. We do expect you to use AI tools like Claude fluently in your own work
• Healthcare or FHIR experience is a strong plus and something you'll go deep in here. If you haven't worked in the domain yet, that's fine; it's learnable, and we'll help you get there