Member of Technical Staff

The hard center of Ren: getting one real-world task done by moving between a live phone call and a browser session, reliably, as concurrency grows. One missed task costs trust permanently, so reliability isn't a chore here. It's the product. T-SHAPED, DEEP IN AT LEAST ONE Agent runtime & orchestration. Deciding when Ren calls, when it clicks, when it stops and asks you. Tool use, model selection, and output quality in production. Telephony & voice. Real-time audio over Twilio and the OpenAI Realtime API, sub-second latency, the difference between a call that sounds human and one that gets hung up on. Reliability & evals. Making "Ren did the thing" true every single time, and proving it with evals instead of vibes. The operational map. Turning every completed task into structured knowledge of how a booking platform, an airline, or a clinic actually behaves. The flywheel that compounds with usage. OUR STACK We work mostly in Node on Firebase Cloud Functions and Cloud Run, with Firestore, Twilio, and real-time model APIs in the loop. Not knowing our exact stack isn't a dealbreaker; depth and ownership are.