2 hours ago
Base Salary
$300k - $450k/yr
Responsibilities
- Design, train, and ship speech-to-speech and real-time voice models for latency, turn-taking, naturalness, robustness, and interruption handling.
- Build and improve production speech infrastructure including neural codecs, streaming inference, ASR/TTS or speech-to-speech pathways, turn detection, and speech enhancement.
- Develop evaluations, harnesses, and monitoring based on real borrower conversations and production quality.
- Diagnose telephony noise, barge-in, accent, channel, and overlapping-speech failures and convert them into model, data, or systems improvements.
- Collaborate across models, data, infrastructure, and product to improve end-to-end voice-agent quality.
- Own major technical areas from problem definition through production deployment.
Requirements
- 5+ years of experience in speech, audio, or related machine learning, or equivalent demonstrated depth through shipped systems and public work.
- Proven experience with speech-to-speech, full-duplex, or other real-time conversational speech systems beyond offline ASR or TTS alone.
- Publicly inspectable work such as papers, GitHub repositories, Hugging Face models or datasets, or similar artifacts.
- Strong engineering fundamentals and experience building reliable streaming or low-latency production systems.
- Ability to diagnose real-world model failures and turn them into practical improvements.
- Experience with neural audio codecs, vocoders, or discrete speech representations is preferred.
- Experience with turn detection, barge-in, speech enhancement, or multilingual speech in production telephony or contact-center settings is preferred.
- Experience adapting language models for voice agents and building evaluations that predict production quality is preferred.
- Familiarity with regulated or high-stakes conversations is preferred.
Benefits
- Fully remote role; the posting also describes collaboration with the Applied AI team at the San Francisco headquarters and contains a conflicting statement that the team typically spends four days per week in person.
- Equity is included in the compensation package, with benefits confirmed based on the hiring country.
- Medical, dental, and vision coverage.
- Generous 401(k).
- Catered lunches.
- Full-time role with a stated typical work schedule of approximately 60 hours per week beginning at 8:00 AM.
Categories
About Salient
Salient builds compliance-first AI agents that automate loan-servicing workflows for US consumer lenders, including collections, customer support, QA/compliance reviews, disputes, and total-loss mitigation. The platform integrates with existing banking and servicing systems and handles borrower communications across voice, text, and email. Founded in 2023 and based in San Francisco, the privately held company sells to banks, credit unions, auto lenders, and fintechs.