A

AMD

Designs chips for computers and AI systems

Lead Functional Validation Engineer - High Speed I/O, PCIe

Full-TimePosted on 8/2/2026Deadline 6/12/27
No salary listed
Senior
Bayan Lepas, Penang, Malaysia
In Person
Hardware EngineeringHardware Validation & Testing

About the job

Requirements
  • Senior-level experience validating high-speed interface controllers, including technical ownership of an interface or subsystem through at least one silicon program from power-on to production release.
  • Thorough working knowledge of PCIe Gen4, Gen5, or Gen6 and coherent socket-to-socket or die-to-die interconnects such as AMD Infinity Fabric (xGMI), UALink, or UCIe at the transaction, data link, and link training layers.
  • Knowledge of link controller configuration and capability space, packet formats and ordering rules, flow control and credit management, replay and retry, link training state machines, power state transitions, and error detection, logging, and containment.
  • Hands-on experience with protocol analyzers, exercisers, and error-injection equipment; register-level debugging through firmware, driver, or JTAG interfaces; and root-cause analysis using captured protocol traces.
  • Proficiency in Python for test content, instrument control, automation, and data analysis, with version control and code review as standard practice.
  • Working knowledge of BIOS/UEFI enumeration and configuration, ACPI, and Linux driver and IOMMU behavior as they relate to link controller operation.
  • Ability to lead cross-functional debug across design, firmware, signal integrity, and platform teams, and communicate findings to engineering and program management audiences.
Responsibilities
  • Develop and execute comprehensive functional validation plans for xGMI, PCIe, and UALink link controllers, ensuring protocol, feature, and performance requirements are met before product release.
  • Validate controller operation at the transaction and data link layers, including configuration and capability space, enumeration, packet generation and handling, ordering rules, flow control and credit management, replay and retry mechanisms, and completion handling.
  • Validate link training and link state management, including link training state machine transitions, recovery and retrain paths, width and speed negotiation, hot-plug and surprise-removal handling, and degraded-mode operation.
  • Validate link power management, including active state power management, device and package state transitions, clock request handling, and entry and exit latency under active traffic.
  • Design and execute protocol-level error injection and verify that error detection, logging, containment, and recovery behave as architected, including error reporting registers, poison generation and propagation, and firmware-first and machine-check flows.
  • Verify register and configuration space correctness, including reset default values, access attributes, and sticky behavior across reset domains, through automated comparison against the register specification.
  • Lead complex debug activities using protocol analyzers, exercisers, register-level tracing, and on-die telemetry to determine root cause and partition issues among controller logic, firmware, physical layer, and platform software.
  • Develop and maintain validation automation, test content, and reporting infrastructure in Python, enabling results to be trended across silicon revisions, firmware builds, and system configurations.
  • Collaborate with design, architecture, and pre-silicon verification teams to align coverage, correlate silicon behavior against simulation and emulation results, and ensure silicon findings are reflected in pre-silicon coverage models.
  • Participate in design reviews and provide input on register maps, error logging definitions, and DFx features to ensure adequate observability and controllability for post-silicon validation.
  • Establish entry and exit criteria for each validation phase and provide functional readiness recommendations at program milestones, supported by documented evidence.
  • Mentor and provide technical direction to validation engineers, and act as the technical escalation point for internal platform teams, ODM and OEM partners, and customers.
  • Document validation methodology and results, and communicate status, risks, and issues to project stakeholders and management.
Desired Qualifications
  • Familiarity with C or SystemVerilog.

About the company

AMD designs processors, graphics chips and AI accelerators for computer makers, cloud operators and businesses that build electronic devices. Its Ryzen processors power personal computers, while EPYC processors handle work in data centers and Radeon chips render graphics. Instinct accelerators run AI models, paired with software that developers use to train models and put them into production. AMD sells chips and related computing systems through equipment makers, distributors and retail channels, and contracts with outside foundries to fabricate its semiconductor designs. It was founded in 1969 and has its headquarters in Santa Clara, California.

Company Size

10,001+

Company Stage

IPO

Headquarters

Santa Clara, California

Founded

1969

Get referred to AMD

See people who can refer or advise you

Simplify Jobs

Simplify's Take

What believers are saying

  • Data centers drive growth: AMD's revenue rose 50% to $11.5 billion in the June 2026 quarter.
  • US learning support is conditional: AMD reimburses up to $5,250 yearly for education, subject to manager approval and funding.
  • Planned demand has an anchor: AMD's July 2026 Anthropic deal targets up to two gigawatts of computing from 2027.

What critics are saying

  • Advanced chip supply relies on one foundry: TSMC makes all AMD processor and graphics wafers at 7nm or smaller.
  • Gaming is shrinking despite AI growth: AMD's gaming revenue fell 31% to $779 million in the June 2026 quarter.
  • Lab work limits location flexibility: AMD lists its Santa Clara optical-interconnect integration engineer role as on-site.

What makes AMD unique

  • Two console families use its designs: AMD designs custom processors for Sony's PlayStation 5 and Microsoft's Xbox Series.
  • Developers inspect its GPU software: AMD's ROCm stack is open source, while Nvidia's CUDA platform remains proprietary.
  • Chip circuitry stays adjustable: AMD sells adaptive processors whose hardware configuration changes after manufacture, without replacing the chip.

Help us improve and share your feedback! Did you find this helpful?

Benefits

Hybrid Work Options

Growth & Insights and Company News

Headcount

6 month growth

2%

1 year growth

3%

2 year growth

-1%
Yahoo Finance
Oct 13th, 2026
AMD shares slip 1.2% to $587 despite Swedish MI355X AI supercomputer plan

AMD shares fell 1.2% to $587 at 10:29 EDT on October 13, despite news of a planned Swedish AI supercomputer built on its MI355X chips. According to Reuters, cloud provider Scaleway plans to offer the system to European businesses and AI developers. Each MI355X accelerator carries 288GB of HBM3e memory and 8TB/s of memory bandwidth. The announcement gave no contract value or chip count, so the revenue impact is unclear. At $587, the shares trade 101.77% above GuruFocus's GF Value estimate of $290.93. The next test is execution: getting the service running, drawing sustained use, and turning that into repeat orders.

Yahoo Finance
Oct 13th, 2026
AMD and SK Hynix could overtake Micron's $1.12T market cap within five years

Micron Technology has become the 11th largest company in the world, with a market cap of $1.12 trillion, driven by demand for memory chips in AI data centres, smartphones, and PCs. One analyst argues two rivals could overtake it within five years. Advanced Micro Devices (AMD), valued at $970 billion, saw second-quarter revenue rise 50% year over year to $11.5 billion. Data centre revenue jumped 107% to $6.7 billion, and AMD's share of the server processor market rose to 34.5% from 27.3%. SK Hynix, valued at $780 billion, holds 50% of the high-bandwidth memory market against Micron's 18%. It trades at 11 times earnings against Micron's 25, and analysts expect long-term earnings-per-share growth of 82%, versus 65% for Micron.

Yahoo Finance
Oct 13th, 2026
Fermi shares jump 13% on AMD credit support deal for 222MW Texas AI site

Fermi Inc shares rose 13.2% in premarket trading Tuesday after the company signed a credit support agreement with Advanced Micro Devices (AMD) for Project Matador, its AI infrastructure site in Carson County, Texas. The agreement backs the lease for phase one, which will deliver 222 megawatts of gross power capacity. It marks a key step toward securing project financing. The lease was announced on August 10, 2026. The companies also signed a framework for possible expansion of up to 650MW. "Our strategic partnership establishes a framework for phases two and three," said chief executive officer Lee McIntire. Fermi has invested more than $1.5 billion in the project so far. Subject to binding customer agreements, the site is expected to scale to about 17GW.

The Register
Oct 12th, 2026
AMD unveils Helion racks for AI agents with up to 32,768 Zen 6 cores

AMD has unveiled Helion, a rack reference design for running AI agents, at the Open Compute Project (OCP) Summit 2026 in San Jose, California. It differs from Helios, the company's rack system for frontier-scale AI training and large-scale inference, which launched earlier this year. Customers can choose a CPU-only hyperscale setup or an enterprise version that mixes CPU and GPU blades. Racks come liquid-cooled or air-cooled, with up to 32,768 "Zen 6" CPU cores, 16 Instinct MI350P GPUs, and 102 Tbps of networking bandwidth. AMD plans to contribute the specifications to OCP. Partner-built systems are due in November 2026 from Cisco, Dell, HPE, Foxconn, Lenovo, Supermicro, and others.

Associated Press
Oct 12th, 2026
QCT unveils AMD EPYC 9006 servers for AI and HPC, including liquid-cooled 8-GPU system

Quanta Cloud Technology (QCT) unveiled two server platforms built on AMD's upcoming EPYC 9006 Series CPUs at the 2026 OCP Global Summit in San Jose on October 12, 2026. The QuantaGrid S46V-1U and D76W-2U target AI, high-performance computing, and cloud workloads. The D76W-2U is a fully liquid-cooled 2U system for large-scale AI. It has dual processors, an 8-GPU baseboard, PCIe Gen6, and up to 800 Gb/s networking. The S46V-1U is a single-socket 1U server for enterprise, cloud, and HPC use. It supports up to four PCIe Gen6 slots and sixteen DDR5 memory channels. The EPYC 9006 chips offer up to 256 cores and 512 threads. Both systems follow open, modular design standards backed by the Open Compute Project.

INACTIVE