Company profile

Zyphra Tech Stack

AI research lab building open-source language, audio and agent models

Technology, Information and Internet San Francisco, California 51–200 employees Privately Held

Zyphra is a San Francisco AI company training its own foundation models, with a stack spanning PyTorch, CUDA, Triton, Slurm, Ray and vLLM/SGLang across AWS, Azure and GCP. Electron, Tauri and Qualcomm/Intel signals point to on-device and desktop delivery, while adoption of RAG, Puppeteer and Playwright aligns with active work on agentic systems and retrieval pipelines. Hiring is engineering- and research-heavy with minimal recent velocity.

Tech Stack 40 technologies

Core StackPyTorch Python Ansible Terraform AWS Docker Kubernetes JavaScript TypeScript React Figma Playwright RAG Hugging Face n8n Azure GCP Apptainer Slurm vLLM Ray SGLang Electron Tauri Puppeteer CUDA Triton Qualcomm Intel Discord+10 more

AdoptingRAG Playwright Puppeteer

What Zyphra Is BuildingChallenges and projects from their job posts

◆Challenges

Reliability and incident response for ML workloads
Scaling compute environments
Optimizing LLM training performance on GPU hardware
Solving fundamental bottlenecks in contemporary models
Growing open-source adoption and community engagement

▲Active Projects

Open-source text-to-speech and audio models
Agentic systems with secure virtualized execution runtimes
Search, retrieval and RAG backend pipelines
Build, deployment and release systems
Observability systems

Hiring Activity

Minimal15 roles · 0 in 30d

Department

Engineering

Research

Sales

Seniority

Mid

Lead

Senior

Principal

Companies like Zyphra

900+ Technology, Information and Internet companies running PyTorch and Python

See the full list

About Zyphra

Zyphra is a privately held AI company headquartered in San Francisco, California, with 51–200 employees. It describes itself as a full-stack AGI company, covering model research, large-scale training, inference and end-user applications. Current work spans open-source text-to-speech and audio models, agentic systems with secure execution runtimes, and retrieval pipelines over structured and unstructured data. The team operates multi-cloud GPU infrastructure and engages a developer community through open-source releases and Hugging Face. Hiring is focused in the United States.

HeadquartersSan Francisco, California

Company Size51–200 employees

Hiring MarketsUnited States

Frequently Asked Questions

What tech stack does Zyphra use?

Zyphra uses PyTorch, CUDA and Triton for training, Slurm, Ray and Kubernetes for orchestration, vLLM and SGLang for inference, and Terraform and Ansible across AWS, Azure and GCP, plus React, TypeScript, Electron and Tauri.

What is Zyphra working on?

Projects include large-scale audio training runs, open-source text-to-speech and audio models, agentic systems with secure virtualized runtimes, RAG backends, and search and retrieval pipelines.

Is Zyphra hiring engineers?

Yes, open roles are concentrated in engineering and research, mostly mid-level, all based in the United States, though recent posting velocity is minimal.

Similar Companies in Technology, Information and Internet

Other companies in the same industry, closest in size

Fluidstack

Technology, Information and Internet

Retell

Technology, Information and Internet

JazzX AI

Technology, Information and Internet

adaption

Technology, Information and Internet

How this profile is built

Zyphra's technology stack, projects, and hiring signals are inferred from public hiring and company data — career pages, public listings, and company web presence — then clustered and de-duplicated. Figures are estimates that refresh over time. Read our full methodology →

This is not an official vendor or customer list. It is a technology-adoption signal inferred from public data, intended for B2B research.