infinity-constellation - Senior AI Platform Engineer - Supernal

• Build advanced AI agent workflows on n8n and Supernal's proprietary platform • Design, implement, and deploy conversational agents, including multi-turn flows, state management, and tool usage • Own end-to-end technical delivery for high-priority customer implementations, from architecture through production launch • Translate customer requirements and SOWs into clear technical designs, execution plans, and deliverables • Make and own architectural decisions across LLM orchestration, RAG design, API integrations, and workflow decomposition • Handle real-world voice system challenges including latency, interruptions, fallbacks, error handling, and failure recovery • Write automated tests — unit tests for isolated logic and end-to-end tests for full workflow and user journey validation • Apply solid error handling: distinguish retryable vs. fatal failures, surface meaningful error messages, and avoid silent failures • Actively debug complex production issues across agent logic, prompts, integrations, and external dependencies • Partner with delivery and product leadership to manage timelines, scope, and technical tradeoffs during implementation • Review technical work for quality, scalability, and maintainability, setting a high bar for engineering excellence • Define, document, and evolve best practices for building and delivering reliable AI Employees • YOU MIGHT BE A FIT IF YOU... • Have 4+ years of experience as a software engineer, automation engineer, or systems builder shipping production systems • Have hands-on experience deploying voice agents using leading platforms (e.g., ElevenLabs, Retell, Nextiva), including telephony and streaming audio integration patterns • Understand multi-turn conversation design: state management, context windows, interruption handling, and graceful recovery • Have tackled real-time constraints in production: latency budgets, streaming audio, fallback paths, and API chaos • Write automated tests as a matter of course — unit tests, integration tests, and end-to-end workflow validation — and treat testing as part of shipping, not an afterthought • Apply solid engineering fundamentals: error handling, retry strategies, separation of concerns, and clean interfaces between components • Are comfortable owning delivery outcomes end-to-end — not just writing code — including timelines, reliability, and client success • Have deep experience with agentic architectures, workflow automation platforms (n8n, Zapier, Make), and APIs • Understand LLM orchestration, prompt engineering, function calling, and retrieval-augmented generation (RAG) • Can prototype fast and finish the job to production quality — with tests, error handling, monitoring, and runbooks • Are an elite debugger who can reason through edge cases, flaky agents, and real-world API failures • Communicate clearly and fluently in English — both in writing and verbally — especially when articulating technical decisions, tradeoffs, and implementation choices to technical and non-technical stakeholders alike • Can run meetings, drive decisions, write crisp updates, and ask the right questions early — without needing heavy process • Thrive in fast-paced, ambiguous startup environments and take ownership without being asked • Bring a low-ego, high-integrity approach to collaboration and leadership • WHAT SUCCESS LOOKS LIKE • Voice-first AI Employees are delivered on time, meet customer requirements, and perform reliably in production • Client implementations are predictable, well-architected, and resilient under real-world conditions • Complex conversational and voice workflows behave consistently and recover gracefully from failure • Code is well-tested, well-documented, and maintainable — not just functional • Technical decisions are communicated clearly and proactively to stakeholders, with tradeoffs explained and risks surfaced early • Engineering best practices reflect real production learnings and are widely adopted across the Mason team • Delivery artifacts — runbooks, SOPs, reusable components — raise the bar for the whole team