--- title: 'Infrastructure Engineer, Factory Acceleration at Etched' canonical: 'https://feeny.ai/job/infrastructure-engineer-factory-acceleration-etched-san-jose-zhzf6k3drvsy' type: 'job' last_seen: '2026-11-02' --- # Infrastructure Engineer, Factory Acceleration at Etched - **Company:** [Etched](https://feeny.ai/companies/etched) - **Location:** San Jose, CA - **Employment:** full-time - **Work type:** onsite - **Posted:** 2026-10-28 - **Last confirmed live:** 2026-11-02 - **Apply:** https://jobs.ashbyhq.com/etched/38d0fd18-2ac1-4825-8182-5d235b266ea3 ## Job description ## About Etched Etched is building hardware for frontier intelligence. We co-design chips, racks, software, and manufacturing to deliver best-in-class throughput and latency across both prefill and decode workloads. Our first products are heavily focused on inference. Backed by hundreds of millions from top-tier investors and staffed by leading engineers, Etched is redefining the infrastructure layer for the fastest growing industry in history. Job Summary Etched is seeking a Test Infrastructure Engineer to build and run the infrastructure behind factory testing of our AI systems, from new product introduction through mass production. You will join a new, small central team that owns making factory testing reliable, repeatable, and scalable. A factory test floor looks a lot like a small datacenter: hundreds of boards, servers, and racks under test at once, all running our test software and reporting results back. You will own that layer: the systems that provision and run test stations, move test results and logs into one reliable source of truth, and roll out new test software safely across factories. You will work closely with hardware, firmware, software, and production teams, and with our contract manufacturing partners. ## Key Responsibilities - Build and operate the test execution platform that runs factory tests across stations, servers, and racks, and keep it running reliably during factory hours. - Provision and manage the Linux fleet on the factory floor: OS imaging and network boot, configuration management, and lifecycle automation for test stations and systems under test. - Design and maintain the networking that test stations and racks depend on, working with factory partners on site. - Build data pipelines that reliably capture every test run, its logs, and hardware genealogy into a central view of test history, yield, throughput, and unit history. - Own versioned release, canary, and rollback of factory test packages so new test software reaches production safely and traceably. - Build monitoring and alerting across factory infrastructure, and lead debugging when infrastructure issues block production. - Integrate with contract manufacturers' factory systems (MES/SFIS) and with shared internal data platforms. - Automate repetitive debug work, such as gathering evidence and grouping repeated failure signatures, and help build toward AI-assisted failure triage. You may be a good fit if you have (Must-have qualifications) - 3+ years of experience building and operating production infrastructure. - Strong Linux administration skills: boot process, package management, systemd, storage, and debugging from the command line. - Solid networking fundamentals: TCP/IP, DNS, DHCP, VLANs, VPNs, and how networks are laid out in server rooms or datacenters. - Experience with distributed systems: services that run across many machines and keep working when some of them fail. - Strong coding skills in Python, C++, or Rust. You build real tools and services, not only configuration. - Experience with automation and configuration management (e.g. Ansible, Terraform) and CI/CD. - Experience with monitoring and observability tools (e.g. Prometheus, Grafana) and debugging production issues. - Bachelor's degree in Computer Science, Computer Engineering, or equivalent practical experience. - Willingness to work onsite in San Jose or Taiwan and travel to manufacturing partners as needed. Strong candidates may also have experience with (Nice-to-have qualifications) - Bare-metal server fleets, lab or datacenter automation, or HPC clusters (e.g. PXE, Slurm). - Server management interfaces such as BMC, IPMI, or Redfish. - Manufacturing test infrastructure, test automation platforms, or factory software (MES, SFIS). - Data pipelines for test results, yield analysis, or hardware genealogy. - AI servers, GPU or accelerator systems, or other high-performance compute platforms. - Working with contract manufacturers; Mandarin is a plus for Taiwan-based candidates. - Bringing first-generation hardware into production at a fast-paced startup. ## Benefits - Competitive compensation packages including generous equity packages - Comprehensive insurance coverage and other top-of-market benefits ## How we’re different Etched believes in the [Bitter Lesson](http://www.incompleteideas.net/IncIdeas/BitterLesson.html). We are the first inference-focused frontier AI system. Our addressable market is the entirety of inference, unlike many of our competitors. We are a fully in-person team in San Jose and Taipei, 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 ## Company Overview - **One-liner**: Etched designs and manufactures specialized ASICs (application-specific integrated circuits) that are hardwired to run transformer-based AI models, offering an order-of-magnitude improvement in inference cost and energy efficiency over general-purpose GPUs. - **Entity Type**: Private (Series B / later stage; total raised $800M as of June 2026) - **Headquarters**: San Jose, California, USA (also has offices in Cupertino, CA; Sacramento, CA; and small presence in Taiwan, Canada, UK, Bulgaria) - **Founded**: 2022 - **Founders**: Gavin Uberti (CEO), Robert Wachen (President & COO), Chris Zhu (Co-Founder) ## Core Business - **Primary industry**: AI hardware / semiconductor design / computer hardware manufacturing - **Target customers**: Frontier AI labs, hyperscalers, and enterprises running large-scale transformer inference workloads (B2B, enterprise) - **Mission / purpose**: Build hardware for superintelligence – specifically, the world’s most powerful servers for transformer inference. ## Products & Services - **Sohu Chip (ASIC)**: A transformer-specific ASIC manufactured by TSMC. It burns the transformer architecture directly into silicon, eliminating the overhead of general-purpose GPUs for inference. - **Frontier Inference Clusters**: Full-stack systems that bundle the Sohu chip with custom-designed racks, networking, and software. Customers place orders for complete systems, not just chips. Etched claims they deliver more tokens per dollar and per watt for dense models, sparse MoEs, diffusion models, and more. - **Software Stack**: Includes compiler and runtime optimizations built by a team with deep experience (former Apache TVM developer, ex-Google TPU software lead). ## Market Standing - **Valuation**: $5 billion post-money valuation (as of December 2025 / disclosed June 2026) - **Key Metric**: $1 billion in contract orders (booked) for its systems; total funding of $800M (including an unannounced $500M round closed Dec 2025 led by Stripes) - **Notable Investors**: Stripes (lead), VentureTech Alliance, Jane Street, Hudson River Trading, Two Sigma, Ribbit Capital. Angel investors include Andrej Karpathy, Geoffrey Hinton, Fei-Fei Li, Arthur Mensch, Scott Wu, Stanley Druckenmiller, Peter Thiel. Earlier backers: Positive Sum, Primary Venture Partners. - **Growth Signals**: - Headcount: 258 employees (as of mid-2026), up 141% year-over-year - 100+ active job postings across engineering, operations, and business functions - $500M raise in late 2025; first chip manufactured by TSMC in 2025 and now being tested with customers - Talent sources include Google (21), Apple (17), NVIDIA (16), Tesla (11), Intel (10), AMD (5) ## Competitive Advantages - **Transformer-specific architecture**: Unlike GPUs (which are general-purpose), Etched’s chip is hardwired only for transformer operations, enabling drastically lower cost and latency for today’s dominant AI model architecture. - **Full-stack system play**: By selling entire inference clusters (chip + rack + software), Etched captures more value and delivers a turnkey solution competitive with NVIDIA’s DGX/HGX systems. - **Deeply experienced leadership**: - CEO Gavin Uberti (Harvard math researcher, AI compiler expert, Apache TVM contributor) - President Robert Wachen (co-founded Prod AI incubator; Thiel Fellow) - VP of Platform (ex-NVIDIA 22 years, built HGX/DGX systems) - VP of Software (ex-Google DeepMind, lead TPU v1–v5 software) - Chief Architect Saptadeep Pal (former Auradine co-founder, chiplet expert) - VP of Hardware Engineering Ajat Hukkoo (ex-Cypress CTO, shipped >$1B revenue products) ## Strategic Focus - **Scale manufacturing and customer deliveries**: With the first chip successfully fabricated, the company is now testing production systems and aims to fulfill $1B in pre-orders. - **Expand engineering talent**: Aggressively hiring across silicon design, validation, infrastructure, optics, and software to support volume ramp. - **Maintain technological lead**: Continue refining the ASIC for future transformer variants and MoE mixtures while building out a partner ecosystem. ## Why Work Here - **Mission-driven, high-impact engineering**: “Building the hardware for superintelligence” – the company is solving one of the hardest problems in compute (inference efficiency) with a focused team. - **Top-tier team and learning environment**: Colleagues from Google, NVIDIA, Apple, Tesla, Broadcom, and Intel; exposure to full-stack chip-to-system design. - **In-office culture**: All roles are in-office (Cupertino and San Jose, CA). The company emphasizes on-site collaboration for hardware development. Notable perks are not heavily advertised, but compensation is competitive for the AI hardware space (equity-heavy). - **Growth trajectory**: 141% headcount growth YoY, hundreds of open roles, and a $5B valuation signal strong backing and career acceleration potential. - **Intellectual challenge**: Work on chip simulation, RTL design, system validation, or software infrastructure for cutting-edge AI inference. ## Sources 1. [etched.com/careers](https://www.etched.com/careers) 2. [builtin.com/company/etched](https://builtin.com/company/etched) 3. [linkedin.com/company/etched-ai](https://www.linkedin.com/company/etched-ai) 4. [cbinsights.com/company/etched](https://www.cbinsights.com/company/etched) 5. [techcrunch.com/2026/06/30/nvidia-competitor-etched-hits-5b-valuation-1b-in-sales-for-ai-chip/](https://techcrunch.com/2026/06/30/nvidia-competitor-etched-hits-5b-valuation-1b-in-sales-for-ai-chip/) ## Other roles at Etched - [Build Engineer](https://feeny.ai/job/build-engineer-etched-san-jose-dmb7by4dkav7) — San Jose, CA - [Inventory Site Lead (Milpitas)](https://feeny.ai/job/inventory-site-lead-milpitas-etched-san-jose-mzfdmk6qz5fp) — San Jose, CA - [Strategic Buyer](https://feeny.ai/job/strategic-buyer-etched-san-jose-d8bna5vet7aw) — San Jose, CA - [People Operations Generalist](https://feeny.ai/job/people-operations-generalist-etched-san-jose-qgrbrkrq0kx2) — San Jose, CA - [Manufacturing Operations Program Manager (Taiwan)](https://feeny.ai/job/manufacturing-operations-program-manager-taiwan-etched-taoyuan-31grtwdgzvxw) — Taoyuan, Taiwan - [Core Engineering Intern](https://feeny.ai/job/core-engineering-intern-etched-san-jose-fcf2ty5ebmnr) — San Jose, CA - [Forward Deployed Systems Engineer](https://feeny.ai/job/forward-deployed-systems-engineer-etched-san-jose-s3zf8j5vd0va) — San Jose, CA - [Electricals Materials Handler (Inventory)](https://feeny.ai/job/electricals-materials-handler-inventory-etched-san-jose-6azxve93tnjp) — San Jose, CA - [Design Engineer, Factory Acceleration](https://feeny.ai/job/design-engineer-factory-acceleration-etched-san-jose-t9g2qtym21s9) — San Jose, CA - [Forward Deployment Engineer](https://feeny.ai/job/forward-deployment-engineer-etched-san-jose-sbpdvtz1r1sb) — San Jose, CA