Deepgram Tech Stack

Real-time speech AI platform for developers and enterprise voice products

Software Development San Francisco, California 51–200 employees Founded 2015 Privately Held

Deepgram operates a real-time voice AI API platform built on Kubernetes, AWS, and NVIDIA infrastructure—processing large-scale audio workloads at low latency. The tech stack reveals heavy investment in distributed compute (Slurm, GPU orchestration) and telecommunications integration (Twilio, WebSockets), paired with active hiring across engineering, sales, and research. Current pain points—GPU cost optimization, scarce training data, latency tuning—signal the company is scaling from API-first adoption toward enterprise reliability and expanding into vertical applications (restaurant automation via the OfOne acquisition).

Tech Stack 24 technologies

Core StackKubernetes AWS Terraform Twilio Cloudflare Salesforce Slack SwiftUI Swift Slurm HPE Dell NVIDIA Groq Linear Notion Swift Package Manager Socket.IO WebSockets iOS

AdoptingOkta Google Workspace JumpCloud

What Deepgram Is Building

◆Challenges

High dimensionality computational storage costs
Scarce audio data
Expensive training deployment
Latency and memory optimization
Customer adoption challenges
High-accuracy menu-aware drive-thru ordering
Real-time analytics and operational intelligence
Increasing top-line revenue
Scaling gpu workloads
Ultra-robust asr in noisy environments

▲Active Projects

Building sales pipeline of new logos
Generating meetings with marketing and sdr teams
Hybrid infrastructure foundation
Website lifecycle management
Ai/ml job scheduling
In-restaurant hardware integration
Real-time analytics and operational intelligence
Automated order-taking platform
Technical discovery for strategic opportunities
Internal data and ml training systems

Hiring Activity

Accelerating65 roles · 35 in 30d

Department

Engineering

Sales

Marketing

Research

Product

Data

Support

Seniority

Senior

Mid

Manager

Staff

Junior

Lead

Director

Company intelligence

Find more companies like Deepgram by tech stack, pain points and active projects

Get started free

About Deepgram

Deepgram provides a real-time speech recognition and conversational AI platform for developers and enterprises building voice-first products. The company serves two primary segments: developer-led adoption (200,000+ developers across 1,300+ organizations using transcription and speech-to-text APIs) and vertical solutions (e.g., drive-thru automation via OfOne). Infrastructure is anchored on GPU-heavy compute (NVIDIA, Groq, Kubernetes, Slurm) to handle sub-100ms latency requirements. Sales and go-to-market are accelerating, with active hiring across engineering, sales, and research teams to expand both platform capabilities and enterprise customer coverage.

HeadquartersSan Francisco, California

Company Size51–200 employees

Founded2015

Hiring MarketsUnited States, Uganda, Indonesia, United Kingdom, Australia, France

Frequently Asked Questions

What is Deepgram's tech stack?

Deepgram uses Kubernetes, AWS, Terraform, Slurm, NVIDIA, Groq, Twilio, and Cloudflare for core infrastructure. Frontend includes Swift/iOS, WebSockets, and Socket.IO. Operational tools: Salesforce, Slack, Linear, Notion. Currently adopting Okta, Google Workspace, and JumpCloud.

What is Deepgram working on?

Active projects include hybrid infrastructure foundation, AI/ML job scheduling, real-time analytics, in-restaurant hardware integration, automated order-taking platforms, and internal ML training systems. Sales and marketing are focused on new logo pipeline and meeting generation.