Nine roles this week. The top pay comes from two speech-model teams. Plaud, maker of the AI note-taking recorder, is starting an in-house speech LLM lab with a $195K–$365K base for founding engineers. Cartesia is paying up to $350K for an applied researcher on its real-time speech-to-speech models. Healthcare voice shows up twice: DrFirst wants an engineer for ambient clinical listening, and Blooming Health is hiring for voice agents in social care. Both are remote. And ServiceNow is hiring a speech linguist with a phonetics background to run language support for its voice platform, a job title you rarely see posted.
Speech & audio research
- Machine Learning Engineer, Speech LLM Training — Plaud
· San Francisco (hybrid, 3+ days in office) · $195K–$365K base + bonus + equity
A founding member of Plaud's new in-house speech LLM lab. You'd build and train large speech models from scratch, whether unified speech LLMs, ASR, expressive TTS or generative audio, and you'd also debug the distributed training clusters they run on. The posting asks for real experience optimizing large PyTorch or JAX training runs. Neural audio codecs, flow matching and RL for conversational timing are pluses. Plaud reports a $300M revenue run rate, profitable without outside funding. View posting - Applied Researcher, Audio — Cartesia
· San Francisco (onsite) · $200K–$350K + equity · visa sponsorship case by case
Research on real-time audio understanding, generation and speech-to-speech models at the company whose founders invented state space models at Stanford. The work covers architectures, multilingual pre- and post-training data (including synthetic data), and how the team evaluates its audio models. Asks for deep expertise in generative modeling and large-scale training. Audio or speech-to-speech experience is preferred but not required. View posting - Sr. Applied Scientist, Alexa Enterprise — Amazon
· Sunnyvale, CA ($192.2K–$260K) or Seattle, WA ($167.1K–$226.1K) · + sign-on + RSUs
Sets the science direction for domain-specific speech-to-text accuracy on a small Alexa team selling to enterprise customers in healthcare, energy, retail and insurance. You'd own model selection and fine-tuning, benchmarks built from real customer audio, and latency across the whole audio pipeline, working directly with customers. Requires a PhD or a master's plus 6+ years. Speech enhancement, beamforming and echo cancellation are listed as pluses. View posting - Senior Audio AI Researcher — Dolby
· Atlanta, GA (hybrid) · $150K–$173K + bonus + equity + profit sharing
Generative, self-supervised and multimodal research in Dolby's Advanced Technology Group, spanning audio, video and text. Examples given include source separation, TTS and music synthesis, with patents and handing technology off to product teams. PhD required, with papers at NeurIPS, ICLR or ICML (ICASSP or ACL is a plus). This is a separate opening from the Multimodal AI Researcher role we listed on September 29, which is still open. View posting
Speech linguistics
- Senior Machine Learning Engineer, Linguistics and Speech — ServiceNow
· Santa Clara, CA (flexible or remote) · $161.3K–$274.2K + equity + incentive pay
A linguist on ServiceNow's voice AI platform. You'd own language support end to end: adding new languages and locales, writing normalization and pronunciation rules, tracing speech failures in production calls back to the responsible component, and designing the experiments that decide which STT and TTS vendors to use. Requires a master's or PhD in linguistics or cognitive science focused on phonetics, acoustics or psycholinguistics, plus 4+ years of applied experiment design and hands-on voice AI work in Python. View posting
Voice agents & voice products
- AI Engineering Lead — TRANZACT
· Fort Lee, NJ · $250K base + annual bonus
A hands-on lead for the conversational AI platform at a Medicare and life-insurance distributor with more than 3,000 employees. That means real-time ASR, agent reasoning and TTS with sub-second turn-taking; SIP, IVR and contact-center integration; conversational RAG; and fine-tuning pipelines judged on WER and latency. Requires 7+ years, at least 3 of them leading production LLM or voice systems. The posting went up in April, so ask the recruiter whether it's still being actively filled. View posting - ML Technical Lead, Agentic Interactions — Sandbar
· New York City (onsite) · $230K–$290K + early-stage equity
Leads the team building the agent inside Stream, Sandbar's private voice ring: conversational interactions, agent control, retrieval, and fast, reliable inference. The founding team comes from Meta, CTRL-labs and Apple. Requires 6+ years in ML, shipped ML products, and LLM evaluation and post-training. Voice ML is a plus. This is consumer product work, not platform or research. View posting - Senior AI Engineer, Conversational AI & Agentic Systems — Blooming Health
· Remote (US) · $170K–$230K + equity
Builds real-time voice agents and the orchestration around them for a platform that connects people to social-care resources: STT, LLM reasoning, tool calling and TTS, plus turn-taking, barge-in and handoffs. "Voice required" is in the job title. You need to have shipped production voice AI on Twilio, WebRTC or similar, and to write strong async Python. SQL and ETL pipelines are also part of the job. View posting - Ambient Listening Engineer — DrFirst
· Remote (US) · $140K–$160K + bonus
Builds the pipeline that captures, transcribes and structures conversations between clinicians and patients, from the microphone through ASR, speaker diarization and noise suppression to structured clinical notes, under HIPAA. Requires 5+ years shipping production systems, streaming audio pipelines, and hands-on work with Whisper, Deepgram or Azure Speech, whether integrating them or fine-tuning your own models. The posting went up August 14. View posting
Hiring for a speech-tech role? Tell us about it and we'll include it here and in the digest. Browse everything by specialty on the homepage.