Software Architect

Acceler8 Talent; Palo Alto (CA); On-site; USD 180,000 - 260,000; Full time

Job summary

Acceler8 Talent is seeking a hands-on Software Architect to take ownership of the accelerator runtime, guiding the project from pre-silicon development to real hardware execution. You will own host-device interfaces, memory management, scheduling and asynchronous execution, working closely with hardware engineers.

You will help optimize AI workloads and lay the software foundation for a completely new AI compute platform, leveraging strong C/C++ and Linux systems expertise in a small,

Qualifications

• Strong C/C++ and Linux systems engineering experience.
• Background in GPU, AI accelerator or HPC runtime development.
• Experience with pre-silicon development, validation or silicon bring-up.
• Deep understanding of hardware/software interaction, memory systems and execution runtimes.
• 4+ years of experience.

Responsibilities

• Architect and build an accelerator runtime from the ground up.
• Own host-device interfaces, memory management, scheduling and asynchronous execution.
• Work directly with hardware engineers on pre-silicon validation and silicon bring-up.
• Optimize AI/ML inference workloads, including data handling and batching.
• Build the software foundation for an entirely new AI compute platform.

Job description

I'm working with an early-stage semiconductor startup building a completely new AI compute architecture, designed to dramatically improve the cost and efficiency of AI inference.

They're developing both custom silicon and the software stack needed to run next-generation AI workloads, with a particular focus on LLMs, agentic AI and physical AI.

With their first-generation chip already taped out, they're looking for a hands-on Software Architect to take ownership of the accelerator runtime and help bring the platform from pre-silicon development through to real hardware execution.

What you'll be working on:

• Architect and build an accelerator runtime from the ground up
• Own host-device interfaces, memory management, scheduling and asynchronous execution
• Work directly with hardware engineers on pre-silicon validation and silicon bring-up
• Optimize LLM inference, including prefill/decode, KV caching and batching
• Build the software foundation for an entirely new AI compute platform

What they're looking for:

• Strong C/C++ and Linux systems engineering experience
• Background in GPU, AI accelerator or HPC runtime development
• Experience with pre-silicon development, validation or silicon bring-up
• Deep understanding of hardware/software interaction, memory systems and execution runtimes
• 4+ years of experience

This is a rare opportunity to join a small, highly technical founding team, work directly alongside the hardware architects and take genuine ownership of a new computing platform.