Shift Left/umc Lab Liaison

AMD AMD · Semiconductors · Austin, TX · Engineering

This role is for a UMC Lab Liaison Engineer at AMD, focused on post-silicon bring-up, debug, and feature enablement for DDR memory subsystems. The engineer will act as a technical point-of-contact for UMC-related issues, own end-to-end execution of memory subsystem validation, and work with LPDDR5/LPDDR5X/DDR5 technologies. The role requires strong debug skills, ownership, and collaboration across validation, firmware, design, and SoC integration teams. While the company mentions AI and data centers, this specific role is focused on hardware memory subsystems and not directly on AI/ML model development or deployment.

What you'd actually do

  1. Own UMC‑related post‑silicon bring‑up, debug, and feature enablement across client and server programs
  2. Serve as the primary liaison between Validation, Firmware, Design, and SoC teams, acting as a technical bridge across functions
  3. Drive end‑to‑end issue ownership by identifying, analyzing, and pushing system‑level memory issues to full resolution
  4. Debug memory subsystem interactions spanning controller, PHY, firmware, and platform components
  5. Perform deep root‑cause analysis using waveforms, lab instrumentation data, logs, and cross‑environment correlation

Skills

Required

  • DDR memory systems (DDR5 / LPDDR5)
  • post-silicon bring-up and validation
  • BIOS, firmware flows, and low-level hardware initialization sequences
  • power states, clocking, resets, and sequencing behavior
  • logic analyzers, oscilloscopes, and silicon debug tools
  • Python or similar scripting languages for automation, log parsing, or data analysis
  • operate effectively in fast-moving bring-up environments
  • communicate across engineering disciplines without direct authority
  • collaborative issue resolution between Design, Firmware, and Validation teams

Nice to have

  • ECC, RAS, memory reliability features, and system-level performance characteristics
  • SoC-level validation, integration teams, and multi-IP subsystem interactions