About the Role Most sales calls don't connect. The ones that do go through a maze: voicemails, phone trees, virtual assistants. Voice AI is the bet that an agent can navigate that maze, guide reps in real time, and turn the resulting transcripts into coaching insights that make whole teams convert better. We run roughly 100 million calls a week today - hyperscaler volume for real-time voice. We're hiring senior engineers to push both sides forward. What you'll do • Build real-time, low-latency Voice AI. Voicemail detection, IVR and virtual assistant navigation, in-call guidance for reps. • Build the call intelligence on top. Transcript analysis, conversation funnels, Markov models of the calls that convert vs. don't, the manager-side product where coaches find patterns across whole teams. • Work directly with customers. Our biggest accounts have shared Slack channels with our engineers; you'll be on calls with users one day and shipping fixes back to them the next. • Make foundational technical decisions. How low can latency go for a voice agent that's still useful? Given the first half of a call, what should the system recommend to maximize the chance the rep books a demo? • Mentor others on the team. Run code and design reviews, and pair on the hardest problems. You might be a strong fit if: • 5+ years of engineering experience, with depth in something close to our stack (React, TypeScript, Node.js, Python). Bonus for experience operating real-time or low-latency systems at scale. • Production experience with AI features. For example, you've shipped a model under a strict latency budget and have opinions about evals. • Strong opinions, held loosely, about what it takes to run real-time voice at hyperscaler volume - capacity, QoS, observability, the things that break first when you 10x traffic. • Not afraid to listen to a few hundred calls to figure out what's actually broken. • You make the rest of the team faster. • Based in San Francisco, excited to be in-office 3+ days per week. At Nooks You'll Enjoy • A world-class team. Around 40 engineers, mostly former founders or founding engineers and veteran builders from places like Applied Intuition, Scale, Airtable, Citadel, and Jane Street. Thirteen Olympiad medals across the team. • A product that drives most of the company's revenue. Teams trial the dialer first and see 5-10x productivity gains within their first two weeks. • Top-of-market compensation.