Deepgram

What is Deepgram?

Deepgram is a voice AI company that builds the real-time speech APIs behind a large share of the emerging "voice AI economy." Its platform spans speech-to-text (STT) transcription, text-to-speech (TTS) synthesis, and full speech-to-speech (STS) capabilities, used by developers and enterprises to power voice agents, call-center analytics, meeting transcription, and real-time human-to-machine interaction. Deepgram was founded in 2015 by three former University of Michigan physicists: Scott Stephenson (CEO), Adam Sypniewski (CTO), and Noah Shutty. The founding insight came from Stephenson's background in particle physics and dark-matter detection, applying deep-learning techniques to raw audio rather than relying on legacy speech-recognition stacks. The company went through Y Combinator and has raised approximately $246M in total. In January 2026 it raised a $130M Series C at a $1.3B valuation, led by AVP, making it one of the newest unicorns in voice AI. Deepgram is headquartered in San Francisco with roughly 326 employees and a remote-friendly workforce. Its core differentiators are speed (real-time, low-latency inference), accuracy on messy real-world audio, and a developer-first API model that competes with the speech offerings of the large cloud providers and newer voice-AI startups.

Mission & values

Power the voice AI economy by giving every developer fast, accurate, and affordable speech understanding and generation, so machines can listen and speak as naturally as people do.

Qualifications

Deepgram hires across engineering (machine learning, speech and audio, NLP, distributed systems, real-time inference infrastructure, backend, frontend, security, SRE), applied AI research, product, design, developer relations, sales, sales engineering, customer success, marketing, and G&A. Application flow lives at deepgram.com/careers. Machine-learning, speech, and real-time-systems roles are the most active given the product runs low-latency inference over live audio at scale. The company is remote-friendly with a San Francisco hub; many US engineering and GTM roles are fully remote, while some leadership and specialized research roles prefer Bay Area proximity. Candidates with production experience in speech recognition, audio ML, model deployment, or real-time streaming systems are strongly prioritized.

Leadership

Hiring process

1. 1

   Apply at deepgram.com/careers

   Browse open roles and submit through the Deepgram careers portal.

2. 2

   Recruiter screen

   Confirms role fit, target level, location preference (remote or SF), and salary expectations.

   About 7 days

3. 3

   Hiring manager interview

   Deep conversation with the manager who owns the position and the team context.

   About 7 days

4. 4

   Technical or role-specific loop

   Engineering and applied AI: coding, ML or speech system design, and often a take-home or live exercise on audio modeling or real-time inference. GTM: case study and role-play.

   About 14 days

5. 5

   Values interview and offer

   Cross-functional interview on collaboration and ownership, followed by an offer with base, bonus, and equity.

   About 7 days

Funding

Awards & recognition

• Series C: $130M at a $1.3B valuation (voice-AI unicorn) · 2026

  Deepgram

• Y Combinator alum building foundational speech AI · 2016

  Y Combinator

Company information

• Industry: Voice AI / Speech Recognition / Developer API
• Location: San Francisco, CA, USA (Remote-Friendly)
• Founded: 2015
• Employees: 326
• Website: deepgram.com ↗
• LinkedIn: Company page ↗
• X / Twitter: Profile ↗

Frequently asked questions