Firmware engineer

We are a new embedded system services company specializing in end-to-end Embedded system design services, SoC design services, FPGA design services with a deep focus on AI/ML offloading SoCs / Physical AI. The founders are a set of Ex-Broadcom, Ex-Intel, Ex-Apple high ranking engineers who already possesses significant experience in embedded system design service industry. The founders’ portfolio includes CXL switch firmware designs and implementations, several designs involving AMD/Xilinx Ultrascale+ MPSoC/RFSoC, Several Custom ASICs, and hardware/firmware designs for multiple physical AI SoC startups in the valley.

This is a founding engineering role within a small, lean, and execution-focused team. You work directly with the CTO as who is an industry veteran with more than 30 years of experience in HW/FW/SW design and implementations across a wide variety of systems. As we grow the team, this would represent a good opportunity for you to move up the management ladder.

During the team building phase, your part-time involvement is expected. Once the company starts full-fledged execution of the projects, you would be offered a full-time employee position with benefits and a portion of the company ownership and profits.

Responsibilities

• Architect Firmware & Drivers: Design and implement low-level firmware, kernel drivers, and application specific interfaces for for AI-specific SoCs.
• Hardware bring-up: Define and implement the Hardware Abstraction Layers and Board Support Packages for various OSs/Bare-metal libraries across a variety of processor architectures (Risc-V/ARM64/x86_64)
• Performance and Footprint Optimization: Manage complex memory hierarchies (SRAM, HBM, DDR) and DMA engines to ensure data flows to the compute cores without bottlenecks.
• Strategic Influence: As a founding member, work closely with the CTO to build the engineering team and set the standard for our development culture.

Requirements

• Deep Systems Knowledge: Expertise in C/C++ and assembly, with a track record of developing for RTOS or embedded Linux
• Fluency in AI/ML: Familiarity of concepts like computational graph, convolution and Transformer layers, Multiply-Accumulate operations on hardware.
• Conversant in high speed interconnects: Good grasp of common high speed interconnects – UCIe, PCIe, CXL ...etc.
• Knowledge in common hardware architectures: Comfortable reading data sheets, understanding PCIe/UCIe/CXL interconnect bottlenecks, and debugging with logic analyzers or JTAG.

Bonus points

• Experience with CUDA, OpenCL, or ROCm performance tuning.
• Knowledge of quantization techniques (INT8/FP8) and their hardware-level implementation.
• Prior experience at a semiconductor startup or an AI lab.
• Familiarity with various models (Deepseek-v3/YoLo-v8/Llama-v4)

Compensation: DoE, and at par with the industry. Total annual compensation will depend on the individual performance and company profits.