Staff Software Engineer - Platform
About the Role
As a Staff Software Engineer on the Platform team, you will own the technical direction of Alpaca's streaming platform - our Redpanda-based event backbone and the user-facing streaming infrastructure. Platform also owns Authentication, Developer Experience and QA, and AI-Native Engineering. Streaming is roughly two-thirds of the job and where we expect you to become the deepest engineer in the company; across the other three you review the designs, hold an informed opinion, and lead the work where they meet streaming.
This is a Staff role, so we are hiring for direction as much as delivery - the engineer who decides what the streaming platform looks like in two years, writes the RFC that gets three teams to agree, and then builds the hard part. We own client-facing and internal services, maintain shared libraries nearly every team depends on, and manage our own deployments. Most of our time goes into Go, but we also write Helm charts, dig through logs and metrics, and get pulled into a wide variety of problems.
Things You Get To Do
Own the architecture of our event streaming platform end to end - Redpanda, the UTS ecosystem, client libraries and user-facing streaming - and set its two-to-three-year direction
Lead the shift from batch to event-driven across Alpaca's systems, working with the teams who own them rather than around them
Design user-facing streaming at scale: connection fan-out, slow consumers, backpressure, ordering and delivery guarantees, and the failure modes that only show up on a volatile market day
Set the standards for event-driven design - schema evolution, topic ownership, replay, idempotency - and own the shared libraries other teams build against, including the cost of versioning and deprecation
Shape AuthN/AuthZ direction across Alpaca and lead the work where identity meets streaming: authorizing real-time subscriptions, token lifecycle on long-lived connections, multi-tenant isolation
Help shape how Alpaca builds with AI - the schemas, contracts and event streams you own are the substrate our agentic development platform runs on
Partner with SRE on chaos testing and game days for the event backbone, and with our QA stream on test tooling and pre-production gates, so resilience is verified rather than assumed
Lead through influence rather than authority, and multiply the engineers around you through mentoring, design review, and handing off the work that grows people
Join the Platform on-call rotation and act as technical lead during incidents affecting shared infrastructure
Who You Are (must-haves)
8+ years of professional engineering experience, including several years running systems at real scale. You write Go, or have deep experience in another strongly-typed language and will pick it up quickly.
You have architected event streaming systems, not just used them - real production depth in Kafka or Redpanda across partitioning, consumer groups, rebalancing, retention and delivery guarantees, and how each breaks under load.
You have delivered realtime data to clients at scale (WebSockets, SSE or gRPC streaming) and handled fan-out, slow consumers and backpressure in production.
You have driven technical direction across team boundaries without formal authority. You can point to the decisions you owned, the RFCs you wrote, and what happened next.
You have built or maintained libraries many teams depend on, and know what backward compatibility and deprecation actually cost.
You have mentored senior engineers and left the teams around you better than you found them.
You have real depth in at least one adjacent area - identity and access, developer tooling and CI, or testing and resilience - and genuine curiosity about the others.
You know Kubernetes beyond the basics: you can write a Helm chart with relative ease, you understand security best practices, and you have job experience or certifications to prove it.
You have worked with relational databases - we use PostgreSQL, but we are not picky as long as the fundamentals are solid.
You have worked at a startup before - you thrive in ambiguous environments, embrace changing priorities, know when to balance technical debt with speed, and proactively tackle whatever needs to get done without relying on pre-existing processes or documentation.
Who You Might Be (nice-to-haves)
You’ve worked with an IDL or binary serialization format (Protobuf, Avro; we use FlatBuffers) and handled schema evolution across many consumers.
You have experience with change data capture (Debezium, PeerDB or similar) and understand how Postgres logical replication works.
You’ve worked on auth systems - OAuth 2.0, OIDC, JWT lifecycle, Keycloak or Cognito - especially around long-lived connections or machine-to-machine access.
You’ve built agentic or AI-native developer tooling, or designed systems meant to be operated by agents rather than only by people.
You’ve run chaos engineering or game days, or built the load and fault-injection testing that makes them worth running.
You’ve done IaC with Argo CD or Terraform.
You have experience with Grafana’s observability stack - Prometheus, Tempo and all the rest. You might have even written a PromQL query for a new alert once or twice.
You’ve worked in fintech or another regulated domain and understand how compliance constraints shape architecture.