d-Matrix Tech Stack

AI inference accelerator hardware and compiler stack for datacenters

Semiconductor Manufacturing Santa Clara, California 51–200 employees Founded 2019 Privately Held

d-Matrix designs custom silicon and system software for AI inference workloads at datacenter scale, using a deep hardware-software stack spanning SystemVerilog, RISC-V, CUDA, PyTorch, and MLIR-based compiler tooling. The hiring profile is nearly 100% engineering-heavy with significant staff and principal seniority, concentrated on chip design (SOC verification, tape-out) and compiler infrastructure — typical of a semiconductor startup executing a multi-year hardware development cycle. Active pain points around inference latency, memory efficiency, and secure datacenter integration reveal the core technical challenges blocking their path to production.

Tech Stack 86 technologies

Core StackPyTorch C++ Python Go MATLAB Linux UVM SystemVerilog C/C++ SystemC vLLM Triton CUDA RISC-V PCIe 5.0 Bash Git Raspberry Pi Zephyr Ubuntu TCP/IP GPIO UART PowerShell GPU TPU Cadence Verilog VHDL Perl+56 more

What d-Matrix Is Building

◆Challenges

Maximizing throughput
Scaling software infrastructure
Predictable project delivery
Reducing inference latency
Tight development window
Complex software/hardware integration
Optimizing inference performance
Enhancing memory utilization and execution efficiency
Integrating secure computing into ai accelerators
Meeting datacenter security requirements

▲Active Projects

Productize ai compute engine software stack
Build compiler infrastructure
Soc verification environments and tape out efforts
Hardware diagnostic tools for novel hardware stack
Ai compute platform focusing on in-memory compute for ai inference in datacenters
Runtime firmware for multiprocessor soc
Scalable mlir compiler for cloud inference
Productizing the sw stack for the d-matrix ai compute engine
Software kernels for next-generation ai hardware
Chip development cycle program management

Hiring Activity

Steady60 roles · 15 in 30d

Department

Engineering

Manufacturing

Research

Security

Seniority

Staff

Senior

Principal

Intern

Director

Manager

Company intelligence

Find more companies like d-Matrix by tech stack, pain points and active projects

Get started free

About d-Matrix

d-Matrix builds Corsair, a specialized compute platform for running large language model inference in datacenters at higher throughput and lower power than general-purpose accelerators. The company operates as a full-stack semiconductor firm: designing custom silicon (verified in UVM/SystemVerilog), developing runtime firmware and kernel software for their multiprocessor systems-on-chip, and building compiler infrastructure (MLIR-based) to optimize inference workloads. Founded in 2019 and based in Santa Clara, the company employs 51–200 people and is actively hiring across engineering roles in the United States, India, Canada, and Australia. Their roadmap includes productizing their software stack, scaling compiler tooling, and integrating security features to meet datacenter deployment requirements.

HeadquartersSanta Clara, California

Company Size51–200 employees

Founded2019

Hiring MarketsUnited States, India, Canada, Australia

Frequently Asked Questions

What is d-Matrix's tech stack?

d-Matrix uses SystemVerilog and UVM for chip design, RISC-V and PCIe 5.0 for hardware architecture, C/C++ and Python for software, PyTorch and vLLM for AI frameworks, and MLIR-based compilers for inference optimization. CAD tools include Cadence for design.

What is d-Matrix working on?

d-Matrix is focused on productizing their AI inference compute engine, building compiler infrastructure (MLIR-based), executing SOC design and tape-out, developing runtime firmware for multiprocessor systems, and creating hardware diagnostic tools for their novel accelerator stack.