DPU RAS and Debug Architect, Infrastructure Silicon

Meta•Austin, TX

1d•$178,000 - $250,000

About The Position

Meta's Infrastructure Silicon organization designs custom silicon that powers our data center infrastructure — SmartNICs/IPUs/DPUs, AI accelerators, and networking ASICs. We are looking for a Data Processing Unit (DPU) RAS and Debug Architect to define the reliability, availability, and serviceability architecture for our DPU ASICs. You'll set FIT-rate targets from the intended usages and deployment models, define how errors are detected, corrected, and reported, and define the trace, debug, and performance-monitoring architecture that supports both post-silicon debug and software and firmware debugging. You'll decide what belongs in hardware versus firmware versus software, define how the hardware presents itself to firmware and software, and work hand in hand with the firmware, driver, and RTL teams - carrying your designs from early path-finding through implementation and silicon bring-up.

Requirements

Bachelor's degree in Computer Science, Computer Engineering, relevant technical field, or equivalent practical experience
8+ years of relevant industry experience architecting RAS and debug/trace architectures and their hardware/software interfaces for NIC/DPU or comparable ASICs
Experience with RAS concepts: FIT-rate estimation and budgeting, failure modes (including silent data corruption), error detection, correction, and containment, and reliability targets for data center deployments. Familiarity with data center reliability, serviceability, and manageability requirements
Experience with error-protection mechanisms - parity, ECC/SECDED, CRC, data poisoning and poison propagation, and lockstep/redundancy techniques - and their area, latency, and power trade-offs
Experience with memory and interface RAS -- LPDDR/DDR RAS (ECC, on-die ECC, error scrubbing, post-package repair) and PCIe RAS (Advanced Error Reporting, ECRC, link error detection and recovery)
Experience with error reporting and handling architectures (ARM and/or x86) - error logging and registers, interrupts, machine-check and AER-style reporting, and escalation to firmware, host software, and platform/BMC management
Experience with on-chip debug and trace architectures -- JTAG and debug access, on-chip trace, breakpoints and watchpoints, and crash and state capture (e.g., ARM CoreSight or comparable)
Experience with performance-monitoring architectures - hardware performance counters and event telemetry - and their use in post-silicon and software/firmware debug
Experience with DFT concepts (scan, MBIST/LBIST, boundary scan) and how they interact with RAS and debug
Experience with processor ISA debug mechanisms (ARM and/or x86) and instruction/execution trace mechanisms such as ETM
Experience with relevant industry standards and specifications, such as OCP (server, RAS, and telemetry/manageability specifications), JEDEC (LPDDR/DDR), and PCIe
Experience driving analysis independently and influencing architectural direction through data

Nice To Haves

PhD in Computer Science, Computer Engineering or Electrical Engineering
15+ years of relevant industry experience architecting RAS and debug/trace architectures and their hardware/software interfaces for NIC/DPU or comparable ASICs
Experience with hardware description languages (e.g., SystemVerilog, VHDL) and simulation environments used in ASIC development flows
Experience designing on-chip trace and debug subsystems (e.g., ARM CoreSight) and the associated post-silicon debug tooling
Experience defining FIT budgets and reliability targets for hyperscale data center silicon, including soft-error rate (SER) analysis and mitigation
Familiarity with functional-safety standards (e.g., ISO 26262) and reliability qualification methods
Experience developing Python-based (or other scripting) automation pipelines for debug, telemetry collection, and data analysis
Experience designing error-reporting and machine-check/AER architectures and the firmware error-handling flows built on them

Responsibilities

Own the RAS and debug architecture for our DPUs - error detection, correction, and reporting, FIT budgeting, and the trace, debug, and telemetry infrastructure - from early path-finding through silicon bring-up
Set FIT-rate targets from the intended usages and deployment models, and budget them across the design - memories, logic, on-chip interfaces, and links
Define the error detection, correction, reporting, and containment architecture -- parity, ECC/SECDED, poisoning and poison propagation, error logging, and how errors are surfaced to firmware, host software, and platform management
Define the trace, debug, and performance-monitoring architecture for post-silicon debug and for software and firmware debugging - on-chip trace, event and counter telemetry, crash and state capture, and the JTAG/debug-access model
Decide what belongs in hardware versus firmware versus software, and define how the hardware presents itself to them - register and programming models, error and interrupt models, and trace/telemetry interfaces
Work closely with design, DV, and PD teams on feature definition and PPA tradeoff; refine architecture to meet the design and PD constraints. Define architecture to maximize DV complexity including defining specific features to help ease DV
Support Design, DV, PD and DFT teams to resolve interface and integration issues as they come up. Support post-silicon bring-up