Sr. Manufacturing Engineer, Design for Manufacturability & Automation, Annapurna AI Systems Manufacturing, Quality and Reliability

Job description

The Annapurna AI Manufacturing, Quality and Reliability (MQR) team is part of AWS Annapurna Labs, building the AI platforms that power the world's largest cloud. We take custom silicon from design through high-volume production across multiple ODM and JDM partners in Asia and North America, at a ramp rate and scale that leaves no room to inspect quality in after the fact.

We are looking a Senior Manufacturing Engineer to own Design for Manufacturability and manufacturing automation across our accelerator card, baseboard, and rack programs.

Your job is to move quality upstream

make the design manufacturable before the drawing is released, and make the line capable of building and verifying it without manual intervention.

You will sit between hardware design, component suppliers, and ODM process engineering. You will review designs for tolerance capability, land pattern and stencil design, connector retention and insertion, part accessibility, handling, and serviceability — and you will hold the line when a design is not buildable at the yield and rate the program requires. You will then own the automation and metrology strategy that makes the resulting process observable: automated inspection, in-line measurement, fixture design, and the data capture behind it.

Key job responsibilities

Design for Manufacturability

- Own DFM, DFA, and DFT review as a gate prior to drawing release and prior to NPI build, across accelerator cards, baseboards, power distribution, interconnect, and rack-level assembly.

- Perform tolerance stack-up and GD&T analysis, and establish tolerances that a partner factory can hold at Cpk ≥ 1.33 at volume — not tolerances that are theoretically correct and practically unachievable.

- Reconcile component supplier specifications against AWS drawings before release, and resolve conflicts where a supplier's requirement is tighter than the drawing allows.

- Design process margin into the product: solder paste volume against land pattern and via structure, insertion force windows, retention features, and clearance for automated handling.

- Provide design feedback on connector retention, contact-level defect modes, fastener and cable access, and serviceability in the rack.

- Build and maintain a versioned DFM design-rule set that carries across product generations, so each program does not relearn the same lessons.

Manufacturing Automation and Metrology

- Own the automated inspection and in-line measurement strategy — SPI, AOI, AXI, ICT, vision, and dimensional or true-position measurement — including whether each system is capable of detecting the defect it is deployed against.

- Drive measurement system analysis across partners: gauge design, gauge R&R, correlation between AWS and supplier measurement, and elimination of methodology differences that make cross-site data non-comparable.

- Specify and deploy automation that removes manual handling and manual data collection from the line, and automate the capture of measurement and process data into SPC.

- Own test fixture and tooling design, calibration, and health management, including detection of fixtures that induce false failures.

- Assess new ODM lines and new sites for process capability before qualification, and define the equipment and process requirements to close gaps.

Program and Partner Execution

- Partner with hardware and silicon design teams during development to influence the design while influence is still cheap.

- Work directly with ODM and JDM process engineering at multiple sites to implement process changes, and verify in production that they held.

- Drive design and process changes through DCN and ECN, including propagation across all build sites.

- Build mechanisms — control plans, capability studies, design-rule checks, runbooks — that continue to work at sites you are not present at.

About the team

Annapurna Labs is a wholly owned subsidiary of AWS, focused on developing custom silicon and servers including the Nitro, Graviton, Inferentia, and Trainium families of processors. Machine Learning Annapurna (MLA) functions as a vertically integrated team including software, firmware, hardware, and silicon design in a single organization. We are the Annapurna AI Servers and Systems organization under MLA, focused on Hardware Development, Software Development, Fleet Ops Systems, and Manufacturing, Quality, and Reliability.

This position is in the Manufacturing, Quality and Reliability team.