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 & parallel chain-of-thought reasoning agents.

Key responsibilities

• Contribute to the architecture and design of the Sohu host software stack
• Implement high-performance, modular code across the complete Etched software stack, consisting of a mix of Rust, C++ and Python.
• Interface with firmware and drivers teams delivering highest-performance HW/SW stack.
• Work with AI model researchers and product-facing teams building out the Etched serving front-end.

Representative projects

• Build scheduling logic for handling continuous batching and real time inference
• Implement inference-time acceleration techniques such as speculative decoding, tree search, KV cache sharing, etc.
• Implement distributed networking primitives for efficient multi-server inference

You may be a good fit if you have

• Experience with C++ and Python
• Familiarity with transformer model architectures and inference serving stacks (vLLM, SGLang, etc.) or experience working in distributed inference/training environments
• Experience working cross-functionally in large software and hardware organizations

Strong candidates may also have

• Experience with Rust
• Familiarity with GPU kernels, the CUDA compilation stack and related tools, or other hardware accelerators
• Understanding of distributed systems, networking, and parallel programming

Benefits

• Full medical, dental, and vision packages, with 100% of premium covered
• 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

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.

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 & parallel chain-of-thought reasoning agents. Etched Labs is the organization within Etched whose mission is to democratize generative AI, pushing the boundaries of what will be possible in a post-Sohu world. 

Key responsibilities

• Propose and conduct novel research to achieve results on Sohu that are unviable on GPUs
• Translate core mathematical operations from the most popular Transformer-based models into maximally performant instruction sequences for Sohu
• Develop deep architectural knowledge informing best-in-the-world software performance on Sohu HW, collaborating with HW architects and designers.
• Co-design and finetune emerging model architectures for highest efficiency on Sohu
• Guide and contribute to the Sohu software stack, performance characterization tools, and runtime abstractions by implementing frontier models using Python and Rust.

Representative projects

• Propose and implement a novel test time compute algorithm that leverages Sohu’s unique capabilities to unlock a product could never be achieved on a typical GPU
• Implement diffusion models on Sohu to achieve GPU-impossible latencies that allow for real-time image generation
• Optimize model instructions and scheduling algorithms to optimize for utilization, latency, throughput, and/or a mix of these metrics. 
• Implement model-specific inference-time acceleration techniques such as speculative decoding, tree search, KV cache sharing, priority scheduling, etc by interacting with the rest of the inference serving stack.

You may be a good fit if you have

• An ML Research background with interests in HW co-design
• Experience with Python, Pytorch, and / or JAX
• Familiarity with transformer model architectures and/or inference serving stacks (vLLM, SGLang, etc.) and/or experience working in distributed inference/training environments
• Experience working cross-functionally in diverse software and hardware organizations

Strong candidates may also have

• ML Systems Research and HW Co-design backgrounds
• Published inference-time compute research and/or efficient ML research
• Experience with Rust
• Familiarity with GPU kernels, the CUDA compilation stack and related tools, or other hardware accelerators

Benefits

• Full medical, dental, and vision packages, with 100% of premium covered
• 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

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.

Etched was founded on the belief that new hardware will be needed to run human- and superhuman-level artificial intelligence. Our ASIC systems deliver orders of magnitude higher performance than conventional AI chips. Etched is actively developing Sohu, an ASIC exclusively for transformers, which will enable products that would be impossible with GPUs, like real-time video generation models and extremely deep chain-of-thought reasoning. And we're bringing this technology to market in months, not years. Sohu is hard-coded for individual model architectures and only supports transformers, but has an order of magnitude more throughput and lower latency than a B200. 

WHAT YOU’LL DO

• Co-Develop cutting-edge software optimized across hardware and software driving the full potential of very large transformer models.
• Solve challenging problems across the full stack, from low-level drivers and APIs to high-level applications and frameworks.
• Contribute to the development of AI solutions that will revolutionize industries like natural language processing, chain of reasoning automation, generative media.

BASIC QUALIFICATIONS:

• Bachelor's degree or equivalent practical experience.
• 3 years of experience in software development, and with data structures/algorithms.
• 3 years of experience developing and testing software on launching products.
• 3 years of experience with firmware design, implementation, and troubleshooting.
• 3 years of experience with embedded architectures, IO technologies (I2C, SPI, PCIe, DRAM etc.), as well as concepts like scalability, fault tolerance, and consistency.

Preferred qualifications:

• Master’s degree Computer Science, or a related technical field.
• Experience resolving customer-related issues.
• Experience solving ambiguous problems.
• Excellent verbal and written communication skills.

Why Etched.ai?

• Be part of a dynamic startup that's changing the landscape of AI.
• Work on the most advanced AI hardware in the world.
• Enjoy a competitive salary and benefits package.
• Make a meaningful contribution to the future of technology.

If you're a talented software engineer who's excited to shape the future of AI, we encourage you to apply!

To Apply:

Please visit our website at etched.ai/careers to submit your resume and cover letter.

Location: Cupertino, California

Do you want to be at the forefront of the AI revolution? Etched.ai is developing groundbreaking hardware that will power the next generation of artificial intelligence. Etched was founded on the belief that new hardware will be needed to run human- and superhuman-level artificial intelligence. Our ASIC systems deliver orders of magnitude higher performance than conventional AI chips. Etched is actively developing Sohu, an ASIC exclusively for transformers, which will enable products that would be impossible with GPUs, like real-time video generation models and extremely deep chain-of-thought reasoning. And we're bringing this technology to market in months, not years. Sohu is hard-coded for individual model architectures and only supports transformers, but has an order of magnitude more throughput and lower latency than a B200. 

We're searching for passionate software engineers to join our fast growing team. You'll have the opportunity to:

• Design cutting-edge software optimizing the performance of our Sohu chip and unlocking the full potential of very large transformer models.
• Co-Develop with a world-class team of hardware and software engineers to push the boundaries of AI technology.
• Solve challenging problems across the full stack, from low-level drivers and APIs to high-level applications and frameworks.
• Review code developed by other engineers and provide feedback to ensure best practices (e.g., style guidelines, checking code in, accuracy, testability, and efficiency).
• Contribute to the development of AI solutions that will revolutionize industries like natural language processing, chain of reasoning automation, generative media.

We're looking for engineers with:

• Strong programming skills in C++, Rust, and/or Python
• Experience with embedded systems, high-performance computing or distributed systems.
• Deep understanding of computer architecture and operating systems
• Passion for AI and a desire to make a real-world impact

Minimum qualifications:

• Bachelor's degree or equivalent practical experience.
• 8 years of experience in software development, and with data structures/algorithms.
• 5 years of experience testing, and launching software products, and 3 years of experience with software design and architecture.
• 5 years of experience with firmware design, implementation, and troubleshooting.
• 5 years of experience with embedded architectures, IO technologies (I2C, SPI, PCIe, DRAM etc.), as well as concepts like scalability, fault tolerance, and consistency.

Preferred qualifications:

• Master’s degree or PhD in Engineering, Computer Science, or a related technical field.
• 3 years of experience in a technical leadership role leading project teams and setting technical direction.
• Experience with Java or Go, with the ability to resolve customer-related issues.
• Experienced in developing solutions to ambiguous problems with significant impact.
• Excellent verbal and written communication skills.

Why Etched.ai?

• Be part of a dynamic startup that's changing the landscape of AI.
• Work on the most advanced AI hardware in the world.
• Enjoy a competitive salary and benefits package.
• Make a meaningful contribution to the future of technology.

If you're a talented software engineer who's excited to shape the future of AI, we encourage you to apply!

To Apply:

Please visit our website at etched.ai/careers to submit your resume and cover letter.

Location: Cupertino, California

Emulation Software Engineer

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 & parallel chain-of-thought reasoning agents.

Key responsibilities

• Oversee SoC bring-up on emulation platforms; diagnose and resolve failing SoC/Processor tests.
• Develop innovative techniques to accelerate pre-silicon validation and software development.
• Provide support for various emulation environments, utilizing advanced emulation techniques including C/C++ DPI transactors, coverage analysis, and in-circuit emulation for high-speed protocols.
• Collaborate closely with teams across Design, DV, Silicon Validation, Performance, and Software, and partner with leading emulation vendors to enhance platform capabilities and troubleshoot complex issues.

Representative projects

• Develop high-performance software to capture debugging signals and create associated tooling to surface valuable insights for users.
• Implement a hybrid emulation environment using custom DPI-based streaming transactors.
• Create highly configurable chip-to-chip network models using emulation-efficient primitives.

You may be a good fit if you have

• Hands-on experience with emulation on platforms such as Palladium, Protium, Veloce, or Zebu, covering design bring-up, build flows, debugging, and performance tuning.
• Strong experience with C/C++ and Linux system development. Proficiency with SystemVerilog and Verilog, including DPI-based interfaces.
• Practical experience with scripting languages (i.e., Python) for automation.

Strong candidates may also have experience with

• Experience working with UVM verification environments.
• Background in design verification, DFT, and testbench modeling.
• Familiarity with waveform debug tools such as Verdi or SimVision.

Benefits

• Full medical, dental, and vision packages, with 100% of premium covered
• 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

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.