Staff MaaS Backend Engineer
You co-own the MaaS system architecture and technical roadmap while setting Go and API standards. You build inference gateways compatible with OpenAI and Anthropic formats, optimize routing and model serving, operate globally distributed services with strict SLOs, enforce authorization and tenant isolation, implement token metering and billing reconciliation, execute safe migrations, and deliver tracing and cost telemetry with strict log hygiene.
Responsibilities
- Co-own the end-to-end MaaS system design
- Author technical decision records and defend technical trade-offs
- Set Go and API standards and align cross-functional teams
- Own API compatibility for OpenAI and Anthropic formats
- Evolve load-aware, model-aware, and prefix-cache-aware routing
- Optimize token throughput, KV cache tiering, and TTFT latency
- Integrate model deployment, LoRA multiplexing, and autoscaling with Kubernetes
- Operate regional inference pools and active-active control planes
- Define and defend service-level objectives
- Enforce authorization, tenant isolation, rate limits, quotas, and prompt-injection controls
- Build exactly-once token metering and billing reconciliation
- Execute incremental platform upgrades with shadow traffic and dual writes
- Deliver request tracing and cost telemetry
- Prevent unauthorized retention of prompts, completions, and PII
Requirements
- 8+ years of backend engineering experience
- 3+ years owning a high-traffic, multi-tenant API platform for paying customers
- Experience with multi-region active-active deployments, caching, backpressure, and performance engineering
- Experience executing zero-downtime brownfield migrations for stateful subsystems
- Hands-on proficiency with Go-based services and production Kubernetes
- Experience with Envoy, GPU-aware scheduling, PostgreSQL, Redis, and Kafka
- Experience with OpenTelemetry and high-cardinality analytics stores
- Understanding of LLM serving, server-sent-event streaming, KV and prefix caching, and TTFT
- Experience building exactly-once metering and billing systems
- Experience designing fail-closed authorization and cross-tenant isolation
Benefits
- Welfare benefits