Топ-оффер в трендовом домене с очень высокой зарплатой и четким планом действий, хотя культура компании намекает на высокий темп и давление.
Высокий темп
Передовой стек
Высокая зарплата
Трендовый домен
Потенциал роста
Прозрачные ожидания
Мэтч & Сопровод
Для мэтча с этой вакансией нужен Plus
Подключить Plus
Описание вакансии
Текст:
Hirify AI/Оригинал
TL;DR
Staff Conversational Designer (AI): Defining persona, turn-taking, repair, confirmation, and pacing for production voice agents with an accent on real-time speech behavior, latency budgets, and measurable conversational quality. Focus on designing endpointing and barge-in behavior, building evals and transcript-review loops, and creating developer guidance for the Voice Agent API.
Location: Remote in the United States, based in Pacific Time; preference for candidates located in San Francisco
Salary: $180,000–$240,000 annual base salary, plus potential equity and bonus; compensation varies by work location, skills, and experience.
Company
Deepgram provides real-time speech-to-text, text-to-speech, and voice-agent APIs and models for developers and organizations building production voice applications.
What you will do
Define and maintain the persona and voice system for Deepgram voice agents across experiences and use cases.
Design turn-taking, endpointing, barge-in, end-of-turn behavior, repair, no-match and no-input handling, and confirmation strategies with ML and Engineering.
Design latency-aware pacing, including brevity, backchanneling, hold behavior, and filler speech within real-time budgets.
Establish conversational-quality evaluations and transcript-review practices that turn production failures into repeatable design improvements.
Build reference agent experiences and developer-facing guidance demonstrating best-practice conversations on the Voice Agent API.
Set conversation-design principles, review standards, and shared vocabulary while partnering with ML, Research, Engineering, and Product Design.
Requirements
Deep experience designing conversational behavior for LLM-based voice agents or assistants, beyond scripted IVR flows.
Fluency with the speech pipeline from ASR through LLM to TTS, including where design decisions are implemented.
Production experience designing turn-taking, interruption, repair, and confirmation patterns.
Experience building quality measurement practices such as evaluations, transcript review, benchmarks, or structured failure-analysis loops.
Exceptional writing craft, including sample dialogs, design guidance, and documentation that others can build against.
Experience influencing Engineering and ML partners without direct authority.
Experience designing for developers or technical users, including APIs, SDKs, and documentation surfaces.
A working AI practice and a clear perspective on where AI tools help and where they mislead.
Expertise and leadership are valued more than a specific number of years of experience.
Nice to have
Experience with a named assistant or production voice-agent platform.
Practice with Wizard of Oz testing and sample-dialog methods.
Hands-on experience with evaluation tooling for LLM or voice quality.
Experience in high-stakes or regulated conversation domains.
Multilingual or cross-locale conversation-design experience.
Background in high-growth B2B companies with self-serve and enterprise motions.
Culture & Benefits
Remote work with Pacific Time alignment; San Francisco location is preferred.
AI use and experimentation are core expectations, including integrating advanced AI tools into daily work.
The role involves rapid change, continuous learning, experimentation, and adaptation rather than highly prescriptive work.
Compensation includes base salary, with equity and bonus offered separately.
Hiring process
Legitimate recruiting communication comes only from an @deepgram.com email address; suspected impersonation should be forwarded to careers@deepgram.com.
Будьте осторожны
Если работодатель просит войти в их систему, используя iCloud/Google, прислать код/пароль, запустить код/ПО, не делайте этого. Это мошенники. Обязательно жмите «Пожаловаться» или пишите в поддержку. Подробнее в гайде →