Voice AI Platform — Powering the Voice AI Economy
The consensus on Deepgram: Choose Deepgram if you want to work on hard speech and audio AI problems with kind, talented people — but expect intensity when the CEO sets the pace.
Founded in 2015 by Scott Stephenson, Adam Sypniewski, Magesh Swaminathan, and Noah Shutty, Deepgram builds the real-time API platform underpinning the voice AI economy. Over 1,300 organizations rely on Deepgram for speech-to-text, text-to-speech, and voice agent capabilities. The culture is deeply technical — researchers and engineers tackle challenging problems across audio and text domains in a collaborative, low-ego environment. After raising $130M Series C at a $1.3B valuation in January 2026, the company is scaling fast while trying to preserve the small-team feel that employees consistently praise.
Real-time speech-to-text, text-to-speech, and voice agent APIs. Powers 1,300+ organizations including enterprises building voice AI applications at scale.
Proprietary speech recognition models trained on massive audio datasets. Research spans audio processing, NLP, and low-latency inference optimization. Try the API →
~260 people with engineering and research as the core. Cross-functional teams working on speech, language, and infrastructure. Researchers tackle problems across audio and text domains with high autonomy.
Explore open roles at Deepgram below.