FuriosaAI Tech Stack

AI inference accelerators and full-stack software for data center deployment

Semiconductor Manufacturing Seoul 51–200 employees Founded 2017 Privately Held

FuriosaAI builds custom silicon and software for AI inference at scale, combining hardware (NPU design), firmware, and Kubernetes-native orchestration. The tech stack is dominated by inference frameworks (TensorFlow, PyTorch, TensorRT-LLM, vLLM, Triton) and cloud infrastructure (Kubernetes, AWS, Azure, GCP), while active projects focus on chip-to-cloud integration—firmware, SDK development, Kubernetes scheduling, and partner hardware onboarding. The engineering-heavy org (30 of 41 hires) and datacenter-scale resource scheduling pain points indicate a company moving past chip design into full-stack systems.

Tech Stack 92 technologies

Core StackKubernetes AWS Rust Python Go C++ TensorFlow PyTorch Jenkins Kafka Linux Azure GCP AWS SQS OpenID JWT OAuth 802.1X PCIe C C/C++ KVM QEMU virtio Windows Keras CUDA Triton TensorRT-LLM vLLM+61 more

AdoptingOperator SDK

What FuriosaAI Is Building

◆Challenges

Operating multi-regional kubernetes clusters
Delivering ai solutions to partners
Strategic collaboration opportunities
Integrating ai chips into kubernetes
Supporting bare-metal and virtualized environments
Reducing inference latency
Improving system throughput
Implementing zero trust security
Resource scheduling at datacenter scale
Efficiency of large-scale evaluation tasks

▲Active Projects

Firmware development for furiosa devices
Customer training
Cloud-native microservice components on kubernetes
Software defined network architecture
Sdk of furiosa npus
Card-level and server-level ai solutions delivery
Hardware partner integration
Kubernetes stack for ai chips
Resource scheduling solution for datacenter scale
Integration of furiosaai npu into kubernetes

Hiring Activity

Decelerating45 roles · 6 in 30d

Department

Engineering

Sales

Support

Security

Design

Research

Seniority

Mid

Senior

Junior

Lead

Notable leadership hires: Art Director

Company intelligence

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

Get started free

About FuriosaAI

FuriosaAI designs data center accelerators and software for AI inference workloads, founded in 2017 by engineers from AMD, Qualcomm, and Samsung. The company builds custom NPUs, accompanying SDKs, and cloud-native deployment tooling, partnering with TSMC, ASUS, SK Hynix, and Samsung on product integration. With over 140 employees across Seoul, Silicon Valley, and Europe, FuriosaAI serves AI teams in enterprise and cloud environments who need efficient, programmable inference hardware. Current hiring is concentrated in engineering roles across the US, South Korea, and Singapore, with a secondary focus on support and sales.

HeadquartersSeoul

Company Size51–200 employees

Founded2017

Hiring MarketsUnited States, South Korea, Singapore

Frequently Asked Questions

What technology does FuriosaAI use to build AI chips?

FuriosaAI uses Rust, C++, and Python for core development, alongside TensorFlow, PyTorch, TensorRT-LLM, and vLLM for inference optimization. Deployment runs on Kubernetes with support for AWS, Azure, and GCP cloud platforms.

Is FuriosaAI hiring engineers?

Yes. FuriosaAI has 30 active engineering roles open across mid-level (26) and senior (13) positions, with recent postings in the US, South Korea, and Singapore. Hiring velocity is currently decelerating.

Where is FuriosaAI headquartered and how many employees?

FuriosaAI is headquartered in Seoul, South Korea, with offices in Silicon Valley and Europe. The company has over 140 employees and 45 total active job openings.