Matcha
Deepgram logo

Deepgram careers: remote jobs, culture and where they hire

The voice AI API platform started by three Michigan physicists, now past $100M ARR and hiring remotely across the US, EMEA and APAC.

Team of 381$215M+ raised, $130M Series C at $1.3B (Jan 2026)Backed by AVP, Madrona, Tiger Global, Wing and Y CombinatorHiring remotely across the United States, Canada, EMEA and APAC

Who backs Deepgram

Deepgram has raised over $215 million. The early money was a $1.8M seed in September 2016 with Y Combinator and Compound, a $12M Series A led by Wing Venture Capital in March 2020, and a $72M Series B in two parts: $25M led by Tiger Global in February 2021 and a $47M extension led by Madrona with Alkeon in November 2022. On January 13, 2026 it announced a $130M Series C at a $1.3 billion valuation led by AVP, with existing investors Alkeon, In-Q-Tel, Madrona, Tiger, Wing, Y Combinator and BlackRock, and new investors including Alumni Ventures, Princeville Capital, Twilio, ServiceNow Ventures, SAP, Citi Ventures, and the endowments of the University of Michigan, Columbia and Stanford. CEO Scott Stephenson told TechCrunch: "We weren't out looking for a raise, though. We had multiple people coming to us." In August 2026 Singapore's EDBI joined as an extension to the Series C, and the same month Deepgram said it had passed $100 million in annual recurring revenue, up from $21.8M of revenue in 2024 (Latka).

What is Deepgram?

Deepgram builds real-time voice AI infrastructure: speech-to-text, text-to-speech and speech-to-speech APIs, plus the runtime for production voice agents. Its models are named Nova (speech recognition), Aura (speech synthesis) and Flux (conversation-native speech), and they run as cloud APIs or as self-hosted and on-premises software for customers with data residency needs. The company was founded in 2015, went through Y Combinator's Winter 2016 batch, and by its own account shipped end-to-end deep learning speech-to-text years before the 2024 voice AI wave.

More than 200,000 developers and 1,400 organizations build on Deepgram, including Twilio, Cloudflare, Sierra, Decagon, Vapi, Granola, IBM and Jack in the Box. The platform has processed over 50,000 years of audio and transcribed more than 1 trillion words. In January 2026 Deepgram acquired OfOne, a YC-backed voice ordering startup for drive-thrus, and turned it into Deepgram for Restaurants, a vertical business unit that sits alongside the core research teams. Headcount grew from about 164 at the end of 2024 to roughly 310 by April 2026 (Latka), and the company was cash-flow positive in 2025.

Who founded Deepgram

Scott Stephenson, CEO and co-founder. Particle physicist who earned his PhD at the University of Michigan, where his research included building a dark matter detector two miles underground and analysing its waveforms with neural networks on FPGAs and GPUs. He left a postdoctoral research position to start Deepgram in 2015, after he and his lab colleagues recorded their own audio around the clock and found no API that could search it.

Adam Sypniewski, CTO and co-founder. PhD in experimental astrophysics and cosmology from the University of Michigan, where he applied machine learning to constrain the dark energy equation of state. Before Deepgram he built AI systems for autonomous vehicles and cybersecurity on DARPA, DoD and DoT contracts. He brought GPU-accelerated end-to-end deep learning to Deepgram and now oversees research, engineering, security, infrastructure and program management.

Noah Shutty, Co-founder and first CTO (2015 to 2017). Met Scott Stephenson as a student in his University of Michigan lab and co-founded Deepgram while finishing a BS in physics and mathematics (2015). He served as CTO through the YC batch and the 2016 seed, left in 2017, completed a PhD in theoretical physics at Stanford in 2022, and is now a researcher on the Quantum Algorithms team at Google Quantum AI.

Where does Deepgram hire remotely?

Deepgram is a remote-first company headquartered in San Francisco, with an office in Ann Arbor, Michigan, a sales office in New York, an Asia-Pacific headquarters in Singapore and an EMEA base in London. It hires remotely across the United States and Canada, in the United Kingdom and the wider EMEA region including the Middle East and North Africa, and in APAC with live roles in Singapore, Sydney and Melbourne and a Pre-Sales Engineer search in India. Most live roles in September 2026 are US remote; some are pinned to Pacific time or to a hub city such as San Francisco or New York.

United States: 44 open rolesAsia-Pacific: 3 open rolesUnited Kingdom: 2 open rolesEurope: 2 open rolesMiddle East & Africa: 2 open rolesCanada: 1 open roleWorldwide: 1 open role

Open remote jobs at Deepgram

Describe your next role, cut the noise

Why Deepgram is exciting to work for

Deepgram is one of the few places to work on the full voice stack at production scale: the research team trains its own speech models, the infrastructure team runs them on Kubernetes across AWS and the company's own bare-metal GPU data centers, and an Applied Engineering team puts them into customers' call centers, drive-thrus and voice agents. Live postings in September 2026 span text-to-speech research, on-device models for earbuds and wearables, inference services, model evaluation, an Arabic-language solutions architect for the Middle East and the restaurant business unit.

Pay is printed on every US posting, with location tiers stated openly, and postings note that US figures are base salary only.

The company is expanding rather than consolidating. It is opening a Voice AI Collaboration Hub in San Francisco for demos, hackathons and briefings, hiring a Global Head of Real Estate and Workplace to open "multiple workspaces and event spaces across the U.S. and internationally", and set up an APAC headquarters in Singapore in August 2026 under a new VP and General Manager for APAC, Sriram Ved.

Deepgram vibe check

Work week
Remote-first with a flexible schedule: the company says people can shift hours for family obligations, continuing education, hobbies or volunteering, and that it does not watch the clock.
Time off
Unlimited PTO, 12 paid US company holidays, paid sick days and 16 weeks of fully paid parental leave.
Office
Nobody is required to go in: there is an office in downtown Ann Arbor, a San Francisco headquarters and a local office in New York for the sales team; the whole company meets once a year at an offsite, most recently in Cabo.
Pace
Glassdoor rates work-life balance 4.2/5 across 40 reviews and the company 4.4/5 overall across 48 (September 2026), with reviewers describing the work as intense and noting it can be hard to find balance at times.
Meetings
A "responsive Slack culture" with regular all-hands, scheduled employee surveys, Employee Resource Groups and personal-interest Slack groups; occasional travel to offsites and team meetings at Deepgram offices is expected for some roles.
Equity
Every posting offers equity, and several roles add a 10% annual bonus on top of base.
Notable
AI enablement workshops are a listed benefit, and a Head of AI Enablement Engineering seat and a People AI & Automation Engineer role exist to build the internal agents and workflows.

What to expect from the culture

Deepgram's careers page lists six pillars: Collaborative, Curious, Grow Together, Do the Right Thing, Think Smart. Act Fast, and Put Customer First. Its 2023 remote-first post by recruiter Winston Bonnheim put it more bluntly: the company values "productivity over politics" and "actual engagement over 'making an appearance'", and "is pretty allergic to micromanagement and clock-watching. 'Do great work, but live your life' is our attitude."

Every 2026 posting carries a "Company Operating Rhythm" block that sets an explicit expectation: "Every team member who works at Deepgram is expected to actively use and experiment with advanced AI tools, and even build your own into your everyday work. We measure how effectively AI is applied to deliver results." The same block warns that "this may not be the right role if you're not excited to experiment, adapt, think on your feet, and learn constantly, or if you're seeking something highly prescriptive with a traditional 9-to-5."

Employees describe ownership and trust. Brent George, Senior Applied Engineer: "There's a strong sense of ownership within Deepgram Engineering. If you see something that can be improved, you're enabled to go solve it." Ingrid Dorai-Rekaa, Staff Product Designer, points to "the empowering environment that emphasizes trust and responsibility."

How to get hired at Deepgram

Deepgram's engineering loop, as third-party interview guides (Dataford, techinterview.org) describe it for 2026, runs recruiter screen, technical screen, a take-home assignment where you build a small application or service, then a virtual onsite of about 2.5 hours that opens with a live extension of your take-home (roughly 90 minutes of adding features or refactoring with an engineer) and closes with a hiring manager conversation. Research Scientist candidates get a one-hour ML discussion with a senior researcher instead of a coding screen. Expect three to five weeks end to end; Glassdoor's reported average is 21 days.

The company's own advice, from CEO Scott Stephenson's blog post on getting a job in deep learning, is that motivation and strong coding matter more than credentials: "A lot of people think you need a PhD or tons of experience to get a job in deep learning, but if you're already a decent engineer, you can pick up the requisite skills and techniques pretty quickly." Show a project with measured results, a public repo and a write-up. Postings across functions screen for what they call an "AI-native operating model": when you hit recurring work, your instinct is to build the system that removes it.

Deepgram hiring FAQ

How much does Deepgram pay?

As of September 2026, US remote engineering bands run $150,000 to $220,000 for Backend Engineer (Inference Services), Applied ML Engineer and Site Reliability Engineer; $180,000 to $240,000 for Senior Software Engineer, Model Evaluation; $179,800 to $246,200 for Senior Frontend and Full-stack Engineers on the Console team; and $219,300 to $276,800 for Staff Developer Experience and Embedded AI Engineers. Product runs $165,000 to $223,000 for Senior PM and $200,000 to $268,000 for Staff PM; a Director of Research, Text-to-Speech is listed at $262,600 to $328,300 in San Francisco, New York and Seattle and $213,000 to $266,300 elsewhere in the United States; Strategic Account Executives are listed at $330,000 to $380,000 on-target earnings; an Inbound SDR at $50,000 to $60,000; and a Senior Pre-Sales Engineer in Singapore at $87,600 to $132,000. Tiered roles pay more in San Francisco, New York and Seattle than in the rest of the United States.

What benefits does Deepgram offer?

Deepgram covers medical, dental and vision, life, short-term and long-term disability insurance, mental health support and an annual wellness stipend. Remote workers get a quarterly productivity stipend and a one-time office upgrade stipend, US employees get a 401(k) with company match and tax savings programs, and everyone gets a learning and education stipend plus conference participation.

What is Deepgram's tech stack?

Deepgram's core inference and on-device work is in Rust, with C and C++ acceptable and Python expected alongside it, and its public SDKs cover Python, JavaScript/TypeScript, Go, Rust, .NET and Java. Infrastructure is Kubernetes on AWS and on-premise bare-metal GPU servers, managed with Terraform, with Slurm integrated for GPU job scheduling. The marketing site is a Next.js application, and the SDKs are developed in the open on GitHub.

Does Deepgram hire junior engineers or interns?

Deepgram lists Handshake among its recruiting channels and runs a Talent Community for future openings, but no internship postings are live in September 2026. The most junior live role is the Inbound Sales Development Representative at $50,000 to $60,000; engineering and research openings are pitched at senior, staff and director level.

How does Deepgram make money?

Deepgram charges for API usage across speech-to-text, text-to-speech and voice agent products, with cloud, self-hosted and on-premises deployment options, and sells through technology ISVs, co-sell partners such as AWS Connect and Cloudflare, and direct enterprise deals. It launched a "Powered by Deepgram" partner program in January 2026, runs a dedicated federal sales team for agencies such as the VA, HHS and DHS, and opened an India endpoint at api.in.deepgram.com in 2026 for lower-latency regional traffic.

Want remote roles like these in your inbox? Tell us what you are looking for.

Describe your next role, cut the noise