Quadric Tech Stack

AI processor IP for on-device inference without code partitioning

Semiconductor Manufacturing Burlingame, CA 51–200 employees Founded 2017 Privately Held

Quadric designs a general-purpose neural processing unit (Chimera GPNPU) that runs AI inference and C++ code on the same chip — eliminating the developer friction of splitting applications across separate processors. The tech stack (LLVM, PyTorch, ONNX Runtime, llama.cpp, CUDA) reveals a compiler-first architecture; active projects on LLVM enhancement, quantization algorithms, and kernel optimization confirm heavy investment in toolchain maturity. Safety certification and licensee acquisition appear as concurrent blockers, suggesting early-stage commercialization challenges typical of semiconductor IP vendors scaling beyond design partners.

Tech Stack 44 technologies

Core StackC++ Python PyTorch TensorFlow Docker Linux Go PostgreSQL Redis JavaScript React Nomad Jira GitHub LLVM GCC UVM C/C++ Assembly vLLM ONNX Runtime TensorFlow Lite CSS HTML5 Git Hugging Face Transformers llama.cpp CUDA DSP Neon+13 more

What Quadric Is Building

◆Challenges

Reducing inference latency
Meeting safety certification requirements
Post series c to ipo
Making product and toolchain accessible via webplatform
Complex software releases
Safety certification compliance
Acquiring new licensees
Optimizing neural network inference
Optimizing neural networks on quadric epu
Iso 26262 certification for npu software

▲Active Projects

Lowering and optimizing neural networks on the quadric epu
Kernel performance optimization
New generation gpnpu test benches
Project feasibility studies
Go-to-market strategy for chimera processors
Quadric kernel library development
Enhance quadric's llvm compiler
Quantization algorithm development for chimera gpnpu
Verification test plans
Webplatform feature development

Hiring Activity

Steady35 roles · 15 in 30d

Department

Engineering

Support

Data

Finance

Marketing

Seniority

Senior

Director

Manager

C-Level

Mid

Lead

Notable leadership hires: Chief Financial Officer

Company intelligence

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

Get started free

About Quadric

Quadric licenses a processor architecture designed to reduce friction in AI inference deployment. The Chimera GPNPU scales from 1 to 864 TOPS and combines scalar, vector, and matrix execution in a single instruction set, allowing developers to write unified C++ applications rather than partitioning code across multiple processors. The company targets semiconductor, edge-device, and automotive customers who need on-device AI with minimal software overhead. Current hiring spans engineering-heavy roles with significant seniority concentration, plus active efforts in kernel library development, compiler optimization, and test infrastructure — all typical of an architecture vendor moving from R&D toward volume licensing and production support.

HeadquartersBurlingame, CA

Company Size51–200 employees

Founded2017

Hiring MarketsIndia, United States, Taiwan, China, Japan, Israel

Frequently Asked Questions

What is Quadric's processor architecture?

Quadric's Chimera GPNPU is a general-purpose neural processing unit that executes AI inference and C++ code without requiring developers to partition applications across separate processors. It scales from 1 to 864 TOPS and supports scalar, vector, and matrix operations.

What tech stack does Quadric use?

Core tools include LLVM, GCC, PyTorch, TensorFlow, ONNX Runtime, llama.cpp, CUDA, and UVM. The emphasis on LLVM and compiler infrastructure reflects Quadric's focus on toolchain accessibility and kernel optimization.