Site Reliability Engineer (SRE)

The role:

• Own production reliability for a real-time platform where uptime and latency ARE the product — voice, desktop, intelligence, and AI combined; an agent mid-call can't wait for a retry.
• First SRE hired immediately (Day 0–14) for production scaling and SLO ownership; a second joins at the start of Phase 3 for 24/7 coverage.
• Pairs with C1 Platform Foundation on observability and tenancy isolation.
• Startup environment: weekly deploys, 1-week sprints, fail fast, move forward — reliability engineering at that speed, not against it.

What you'll own:

• SLOs and error budgets per tenant/service
• Incident response and blameless postmortems
• Production scaling and capacity
• Observability depth (p50/p95/p99 per event hop)
• Uptime as a personal mission
• On-call rotation with DevOps
• Your committed timelines.

Who you are:

• Self-starter, grit, show-me mentality — you prove reliability with dashboards and drills, not assertions.
• A ways-to-YES engineer: weekly deploys are the heartbeat and your job is making them safe, never slowing them.
• You love new technology, adapt fast when the stack changes under you, use AI tools daily to multiply velocity, and consider yourself exceptional.
• Calm in an incident, relentless after it.
• Team player who likes winning.
• 8+ years operating production systems at scale; owns SLOs, error budgets, incident command.
• Strong Go or Python — you automate reliability, you don't toil at it. Everything you build is code: runbooks execute, remediation is automatic, toil trends to zero.
• Deep on event-driven and real-time systems reliability — NATS-class buses, WebSocket fleets, streaming pipelines — and the failure physics underneath: state, race conditions, locking, ordering, back-pressure, cascading load. You've debugged these in production.
• Strong monitoring and uptime mindset — metrics, logs, traces wired to alerting that catches it before the customer does; you know the difference between a noisy alert and a real signal.
• Good networking understanding — protocols and how they work (TCP/UDP, TLS, WebSocket, DNS, load balancing); RTP/SIP a strong plus for our media paths.
• GCP at scale; multi-cloud literacy a plus. Multi-tenancy isolation experience a strong plus.
• Capacity modeling and load testing partnership with QA — find the knee of the curve before customers do.
• Chaos engineering — failure injection as routine practice; prove graceful degradation, don't assume it.
• Deploy-safety partnership with DevOps — canary analysis, automatic rollback triggers, error-budget-driven release gates.
• AI-aware reliability — monitoring model latency, drift, and cost as production signals, not just CPU and memory.
• Incident communication craft — clear, fast, blameless; execs and customers get truth at the right altitude.
• A master debugger of production — reads the trace, the metric, the flame graph, and sees it; narrows an incident to the service, the deploy, the event.

What You Will Bring:

• 8+ years operating production systems at scale; owns SLOs, error budgets, incident command.
• Strong Go or Python — you automate reliability, you don't toil at it. Everything you build is code: runbooks execute, remediation is automatic, toil trends to zero.
• Deep on event-driven and real-time systems reliability — NATS-class buses, WebSocket fleets, streaming pipelines — and the failure physics underneath: state, race conditions, locking, ordering, back-pressure, cascading load. You've debugged these in production.
• Strong monitoring and uptime mindset — metrics, logs, traces wired to alerting that catches it before the customer does; you know the difference between a noisy alert and a real signal.
• Good networking understanding — protocols and how they work (TCP/UDP, TLS, WebSocket, DNS, load balancing); RTP/SIP a strong plus for our media paths.
• GCP at scale; multi-cloud literacy a plus. Multi-tenancy isolation experience a strong plus.
• Capacity modeling and load testing partnership with QA — find the knee of the curve before customers do.
• Chaos engineering — failure injection as routine practice; prove graceful degradation, don't assume it.
• Deploy-safety partnership with DevOps — canary analysis, automatic rollback triggers, error-budget-driven release gates.
• AI-aware reliability — monitoring model latency, drift, and cost as production signals, not just CPU and memory.
• Incident communication craft — clear, fast, blameless; execs and customers get truth at the right altitude.
• A master debugger of production — reads the trace, the metric, the flame graph, and sees it; narrows an incident to the service, the deploy, the event.