Twenty-one roles this week. The best pay you can earn remotely is at Deepgram, which is paying $216K–$338K for US-remote researchers who take a customer's voice-AI problem, fix it in the models, and ship the fix within weeks. Amazon's Fauna Robotics is hiring three audio specialists at once, a principal scientist, an acoustic engineer and an embedded DSP lead, to work out how a home robot hears over its own motors. Otter.ai wants a director of AI research for ASR, speaker identification and audio language models at up to $313K. If you're early in your career, look at Plaud in Singapore and Audible in Culver City: both ask for just a year of experience. Cartesia is also hiring a contractor to turn voice-actor recordings into its TTS voices.
Speech & audio research
- Research Staff, Applied Voice AI — Deepgram
· Remote (US), San Francisco or New York · $216K–$338K + bonus + equity
Finds where Deepgram's speech-to-text and TTS models fall short for customers and fixes it in production within weeks. The posting's examples include new languages, TTS pronunciation of names and numbers, isolating a speaker from crosstalk, and steerable and cloned voices. You'd set the targets, design evals, curate or generate data and fine-tune. Requires deep experience training and evaluating deep learning models, strong PyTorch, and ML you've shipped to production. STT, TTS or voice-agent experience is a plus, not a requirement. View posting - Research Engineer, Post-Training — Deepgram
· Remote (US), San Francisco or New York · $216K–$338K + bonus + equity
A second opening on the same team, with the same description and pay as the role above but an engineering title. Apply to whichever title fits your background. View posting - Principal Applied Scientist, Real-Time Conversational AI — Amazon (AGI)
· Sunnyvale, CA ($228.7K–$309.4K) or Boston or Seattle ($198.9K–$269K) · + sign-on + RSUs
Sets the technical direction for real-time speech and audio foundation models, along with the reward modeling and RL that shape natural conversational behavior. This is the principal-level version of the senior role we listed last week, posted October 28. Requires a PhD, 10+ years of applied research, and hands-on work scaling foundation models that process or generate speech, text and audio together. View posting - Director of AI Research — Otter.ai
· Mountain View, CA · $244K–$313K
A hands-on research leader for ASR, speaker identification, LLMs and audio language models. You'd own the whole model lifecycle, from data and pre-training to evaluation and deployment, and build the team. Requires a master's or higher (PhD preferred), 10+ years in software or AI research including 4+ managing technical teams, publications and deployed systems, and deep expertise in both speech and LLMs. View posting - Research Scientist II, Deepfake Detection & Authentication (Video & Audio) — Pindrop
· Remote (US) · $160K–$185K + RSUs
Researches how to detect synthetic speech and manipulated video, plus voice and identity authentication, at a voice-security company that says it analyzes more than 8 billion interactions a year. You'd build reproducible benchmarking pipelines, probe models for attack vectors and detection gaps, and hand prototypes to engineering. Requires a master's or PhD (or equivalent research experience), 3+ years of deep-learning research in speech, vision or voice biometrics, and hands-on deepfake detection work backed by papers or patents. ASVspoof or NIST SRE experience is a plus. View posting - Research Scientist (all levels) — Pindrop
· Remote (US) · compensation not listed
Pindrop's general research posting, open at every level from 1–3 years of experience to senior and principal (6+). It covers fraud, voice and speech, video, authentication and deepfake detection. Requires a master's or PhD (or equivalent research experience), strong PyTorch or TensorFlow, and a record of delivered or published research. Apply here if you're earlier in your career than the Research Scientist II role above. View posting - Research Staff, LLMs — Deepgram
· Remote (EMEA) · compensation not listed
An LLM specialist for Deepgram's research team, working on the hard parts: data curation, large-scale distributed training, transformer architecture optimization and RL. The team's stated goal is new ways to build voice AI that get around audio's scarcity and cost problems. One of the few listings this week open to people based in Europe, the Middle East or Africa. View posting
Speech data & ML engineering
- Software Engineer, Multimedia & Multimodal AI — Meta
· Menlo Park, CA (one other location not named) · compensation not listed
Builds the data and evaluation behind Meta's speech, sound and music models in its Applied AI org. The work spans audio corpus pipelines with pseudo-labeling and provenance tracking, objective metrics like speaker and style similarity tied to human listening tests, and training generative audio models directly. Requires 6+ years of programming (or 3+ with a PhD), 3+ years building ML systems, strong PyTorch, and hands-on ASR, TTS, audio codec or generative audio work. View posting - Research Engineer, Data Infrastructure — ElevenLabs
· Remote (UK, US, Poland or Bulgaria) · compensation not listed
Owns the data systems behind ElevenLabs' audio models: large-scale pipelines for collecting, filtering and transforming training data; the classifiers and quality filters inside those pipelines; and the curation choices (deduplication, quality scoring, labeling) that measurably improve the models. No degree required. ElevenLabs asks for evidence like past projects or GitHub work, plus experience with distributed data processing at scale. View posting - Machine Learning Engineer (Speech/Audio) — Plaud
· Singapore (onsite) · compensation not listed
Runs the speech data pipelines behind Plaud's AI note-taker: data collection, cleaning, labeling and augmentation; mining hotwords from ASR output; and fine-tuning for code-switching, names and product terms. You'd also build the eval sets that compare Plaud's models with open-source and commercial ones. Requires just 1+ year in speech, ML or large-scale data engineering, solid Python and PyTorch, and Spark or Ray. View posting
Audio hardware, acoustics & devices
- Principal Audio Scientist, Fauna — Amazon
· New York, NY · $218.8K–$295.9K + sign-on + RSUs
The technical authority on how Fauna Robotics' robots hear. You'd suppress noise from the robot's own motors and fans, cancel echo so people can interrupt it mid-sentence, locate and track people as they move and talk, and make wake word and voice activity detection robust. You'd also drive microphone placement and acoustic hardware decisions. Requires a PhD plus 7+ years (or a master's plus 12+), deep beamforming and echo cancellation expertise, and audio front-end algorithms you've shipped on a physical device. View posting - Senior Acoustic Engineer, Fauna — Amazon
· New York, NY · $157.1K–$212.6K + sign-on + RSUs
Owns the acoustic hardware for Fauna's robots: microphone arrays, speakers, noise and vibration isolation, acoustic simulation in COMSOL, and the measurement fixtures used through production ramp. Requires 7+ years of acoustic design and validation for hardware products, SoundCheck or Audio Precision, and experience fitting transducers into cramped, mechanically complex enclosures. Beamforming or direction-of-arrival experience is a plus. View posting - Sr. Embedded Systems Engineer, Audio DSP, Fauna — Amazon
· New York, NY · $184.9K–$250.2K + sign-on + RSUs
Leads the embedded audio signal-processing pipeline for Fauna's robots: designing DSP algorithms, getting them running on real-time hardware, and tuning them as the hardware changes. You'd also manage outside acoustics consultants. Requires 7+ years developing DSP for embedded or real-time systems, C/C++ plus Python or MATLAB, and experience as a tech lead. Keyword spotting and direction-of-arrival work are pluses. Posted October 8. View posting - Audio Systems Engineer — Sesame
· San Francisco or Bellevue, WA · $175K–$280K
Defines, tunes and ships the speech playback and spatial audio capture for Sesame's compact smart wearable, the hardware for its voice agents. Requires 4+ years in audio hardware with at least one shipped consumer electronics product, hands-on microphone array tuning, SoundCheck or Audio Precision testing, DSP fundamentals (filters, beamforming, echo cancellation) and COMSOL. View posting - Director, Intelligent Audio — HARMAN
· Northridge, CA · $173.3K–$254.1K + bonus
A hands-on lead for a small applied research team that builds AI-driven audio for JBL, Harman Kardon and HARMAN's automotive products, from research through to real-time, embedded implementations. Requires 8–10+ years in audio technology and AI for audio, novel technology shipped in products, and experience leading technical teams. Speech pickup or generation and beamforming are listed among the useful backgrounds. This is a separate opening from the Audio ML Engineer role we listed on October 6. View posting
Voice products & voice production
- Software Development Engineer, Audible AI Voice Production — Amazon
· Culver City, CA ($122.6K–$170K) or Cambridge, MA ($110.5K–$160K, separate posting) · + sign-on + RSUs
Builds tools that automate Audible's audiobook production pipeline for narrators, voice actors, producers and publishers, at a new Audible hub in Culver City. Requires 1+ year of professional software development. A matching Cambridge, MA opening was posted October 15. View posting - Software Engineer, Voice AI / Backend — Clera
· Berlin (onsite) · €85K–€160K + equity · visa sponsorship and relocation
Builds voice AI agents for customer-facing workflows at a small engineering team. That means RAG, dynamic model routing and transcription evaluation, plus a distributed call scheduler that handles concurrency, retries and provider limits. Requires production distributed-systems experience in Python, FastAPI or TypeScript, plus real-time audio or voice infrastructure and production LLM work. LiveKit and Temporal are pluses. View posting - Product and Research Operations Manager, Voices — Cartesia
· San Francisco (onsite) · $160K–$225K + equity
Runs the operations behind Cartesia's voice library: voice production runs for customers, sourcing and vetting voice-acting talent, library health reporting, and delivering voices to sales and customer success. You'd work alongside the Voice Production and Voice Library leads. It's an operations role, not engineering. View posting - Audio Production Expert (Contract) — Cartesia
· San Francisco (onsite) · $45–$60/hour
Turns raw voice-actor recordings into the voices in Cartesia's TTS library. You'd set casting requirements, review recordings as they come in, and build voices to customer briefs (one example: "warm, trustworthy, a little Southern, for a healthcare line"). You'd also help with audio for TTS and ASR research. It's for someone with years of recording and audio engineering experience and a good ear. View posting
Telephony & sales engineering
- Voice Support Engineer — Assort Health
· San Francisco (hybrid) · compensation not listed
Keeps the phone side of Assort's healthcare voice agents working, by diagnosing one-way audio, codec mismatches, registration failures and DID routing problems across clinics' PBXs and carriers. Requires 3+ years supporting VoIP, SIP or enterprise telephony, solid SIP/SDP/RTP knowledge, and Wireshark and PBX logs (Asterisk, Cisco, Genesys). Twilio or SignalWire experience is a plus. View posting - Staff Sales Engineer, Voice AI — Deepgram
· Remote (US) · $185K–$230K + commission + equity
The technical lead on Deepgram's voice agent deals. You'd run discovery, architecture and proofs of concept, and get customers into production. Buyers include contact center, healthcare and financial services teams, and deals turn on latency, reliability and cost per minute. You'd carry a quota for new Voice Agent customers and report to the VP of Sales. 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.