About Etched

Etched is building AI chips that are hard-coded for individual model architectures. Our first product (Sohu) only supports transformers, but has an order of magnitude more throughput and lower latency than a B200. With Etched ASICs, you can build products that would be impossible with GPUs, like real-time video generation models and extremely deep chain-of-thought reasoning.

Firmware Engineer, Machine Learning Accelerator

We are looking for a Firmware Engineer with experience with machine learning accelerators. The ideal candidate will be able to contribute to firmware design, implementation, and performance optimization, as well as debug both firmware and hardware in simulation, emulation, and silicon environments.

Representative projects:

• Develop firmware for machine learning accelerator blocks (e.g. systolic array).
• Work with hardware and system architects to optimize performance of machine learning accelerators using hardware software co-design.
• Collaborate closely with hardware engineers to comprehend architecture and specifications.
• Conduct debugging, testing and validation of machine learning accelerator blocks in simulation, emulation and silicon environments.
• Propose feature/debuggability/performance enhancements for future generations of machine learning accelerator blocks.

You may be a good fit if you have:

• 2+ years of experience in firmware development for RTOS embedded systems.
• 1+ years of experience with machine learning algorithms, especially transformers.
• Ability to learn quickly and have an open-minded mindset.
• Proficiency in C/C++ programming for embedded systems.
• Experience with production silicon and system lifecycle.

Strong candidates may also have experience with:

• Experience working with simulation and emulation environments.
• Experience with ARM/ARC/Xtensa/RISC-V embedded architecture.
• Experience with hardware validation and debugging tools (e.g. OpenOCD/Segger/Lauterbach).
• Experience with electronics lab equipment (e.g. oscilloscopes and protocol analyzers)
• Experience with SystemVerilog language and debugging with waveforms.
• Experience with Python language in scripting and automation.
• Experience with Bazel/Blaze/Buck build system.

How we’re different:

Etched believes in the Bitter Lesson. We think most of the progress in the AI field has come from using more FLOPs to train and run models, and the best way to get more FLOPs is to build model-specific hardware. Larger and larger training runs encourage companies to consolidate around fewer model architectures, which creates a market for single-model ASICs.

We are a fully in-person team in Cupertino, and greatly value engineering skills. We do not have boundaries between engineering and research, and we expect all of our technical staff to contribute to both as needed.

Benefits:

• Full medical, dental, and vision packages, with 100% of premium covered, 90% for dependents
• Housing subsidy of $2,000/month for those living within walking distance of the office
• Daily lunch and dinner in our office
• Relocation support for those moving to Cupertino