Staff SRE Engineer - voice Core Platform

We are building a real-time conversational AI voice platform. An orchestrator runs every call through Voice Activity Detection, Speech-to-Text, LLM-based agent matching, tool calling against ServiceNow platform (Glide) APIs, and Text-to-Speech. In voice, latency and reliability are the product. You will own the platform's reliability, performance and operational architecture, and set direction across teams.

Responsibilities:

• Define and drive SLIs/SLOs for availability, perceived latency (end-of-speech to first audio), and call quality.
• Instrument end-to-end tracing across every pipeline stage, and lead architectural work to cut latency.
• Design autoscaling and capacity plans for stateful, real-time connections (WebSocket, WebRTC, SIP/RTP).
• Lead load, soak and breakpoint testing, and turn results into capacity models.
• Build resilience for LLM, STT and TTS dependencies: failover, fallbacks, rate-limit handling and graceful degradation.
• Own Kubernetes architecture, infrastructure-as-code and CI/CD with safe progressive delivery.
• Lead incident response and postmortems, and mentor engineers across the team.

Requirements:

• Experience in leveraging or critically thinking about how to integrate AI into work processes, decision-making, or problem-solving.
• This may include using AI-powered tools, automating workflows, analysing AI-driven insights, or exploring AI's potential impact on the function or industry.
• 8+ years in SRE, DevOps, production engineering or software engineering on production systems at scale.
• 1+ year building or operating real-time voice systems (VoIP, SIP/RTP, WebRTC, or STT/TTS integration).
• 2+ years building or operating AI agentic systems and orchestration pipelines (agent routing, tool calling, multi-step LLM workflows).
• 1+ year deploying and operating LLMs in production (model serving, inference optimisation, provider integration and failover).
• Strong expertise in Kubernetes, cloud infrastructure ([AWS / Azure / GCP]) and Terraform or similar.
• Hands-on observability experience (OpenTelemetry, Prometheus, Grafana, Datadog or similar).
• Proficiency in Python, Go or Java.

Nice to have:

• Experience in inference serving (vLLM, Triton, GPU scheduling).
• Familiarity with the ServiceNow platform or enterprise SaaS backends.