Confidential client — a Seed-stage startup (1-10 employees) building the simulation, observability, and self-improvement layer for Voice AI agents, with a mission to be the trust layer between AI and humanity. The founders previously built critical infrastructure at major cloud and AI companies. The company builds the evaluation infrastructure to make voice agents reliable enough to succeed with real customers. In just one year, large technology, healthcare, banking, and food-service organizations across five continents have used the product to understand and improve their Voice AI agents. Raised a $4M seed, runs comfortably on seven-figure revenue, and handles millions of calls a month. Team of 10, based in a San Francisco office. Full-Time, In-person, San Francisco, CA. Experience: 1+ years. Salary: $160,000-$220,000/yr. Visa sponsorship: None available. About the Role: As a Member of Technical Staff, you'll architect and develop systems that simulate, analyze, and evaluate conversational AI agents across voice and multimodal systems, and design resilient, scalable infrastructure on AWS (Lambda, EC2, WebSockets, SIP) to handle millions of real-time conversations. You'll build algorithms and pipelines that surface insights, detect failure modes, and keep agents safe, reliable, and aligned with human intent. This is a high-bar, full-ownership role on a 10-person team, with proficiency expected across the stack, especially databases, AWS, and AI engineering. It's also customer-facing: you'll work on-site with customers, building, selling, and deploying the product to some of the largest companies in the world. The company wants engineers who obsess over details, can lead a project's technical execution, and have built high-traffic features at scale. What You'll Own: - Architecting and developing simulation, analysis, and evaluation systems for conversational AI (voice and multimodal) - Resilient, scalable AWS infrastructure (Lambda, EC2, WebSockets, SIP) for millions of real-time conversations - Algorithms and pipelines that surface insights, detect failure modes, and keep agents safe and aligned - Full ownership of your work, across the stack Company stage: Seed. Work type: In-person (San Francisco, CA). No visa sponsorship available.