AI Voice Engineer

I'm building real-time AI voice into a commercial product and need an engineer who has actually shipped live voice agents that place and handle phone calls — not someone who's read about it. Short, sharp engagement now, with real scope to grow into a larger build for the right person. I'm deliberately not naming the tools, models, or providers I'm using. If you're as good as your profile claims, you'll already know the landscape cold and you'll tell me what you'd reach for and why. I'm not here to hand you the answer sheet — I want to see how you think. What I'm actually screening for: • You've put live voice agents into production. Real calls, real users, things that broke and you fixed them. • You can diagnose, not guess — isolate where a pipeline fails before touching settings. Method over flailing. • You understand the full chain end to end: telephony, speech-in, the model, speech-out, latency, turn-taking. Weakness anywhere shows. • You can handle multilingual / non-English voice work and know why it's harder than English. • You move fast, communicate in plain terms, and don't disappear. Don't apply if: you're an agency fronting juniors, your "voice" experience is text chatbots, you need to research before you can speak to it, or your plan is to learn on my time and budget. Your first message — this is the filter: Don't send a proposal. Send me three things, briefly: 1. A live voice agent you built — what it did, what was hard, what broke. 2. A failure you diagnosed where the obvious cause was wrong — and how you found the real one. 3. The stack you'd choose for a production phone-based voice agent today, and why that over the alternatives. Anything generic, anything that ignores these three, anything that reads like a template — auto-rejected. Impress me in the first message or there won't be a second.