Senior SoC Chiplet Architect Job Santa Clara area,California USA,Manufacturing / Production

Job Description:

The CEG NAG (Networking Architecture Group) is ••••• 's premier team focused on defining the future of high-performance networking silicon. Our team architects next-generation networking solutions that enable hyperscale data centers, cloud infrastructure, and AI workloads to achieve unprecedented performance and efficiency. We specialize in IPU/DPU platforms, advanced packet processing architectures, and programmable networking technologies that form the backbone of modern distributed computing systems.

We are seeking a Senior SoC Chiplet Architect to define and lead the architecture strategy for multi-generation, chiplet-based SoC platforms targeting next-generation data center workloads.

This role is responsible for chiplet partitioning, die-to-die (D2D) interconnect architecture, and system-level tradeoff analysis across performance, power, area (PPA), cost/yield, and software complexity. You will drive modular, scalable SoC design approaches, including cross-chiplet coherency, system infrastructure (boot/reset/telemetry), and RAS/security architectures across multi-die systems.

Key responsibilities will include but not limited to:

Chiplet Architecture Strategy and Roadmap

Define the monolithic vs. chiplet decision framework and multi‑generation roadmap, incorporating reticle scaling limits, yield economics, modularity, and portfolio reuse strategy

Lead architecture studies for 2‑die / 3‑die / multi‑chiplet scaling (compute + I/O/network + memory/accelerators), including partition boundaries, SKU scalability, and future‑proof modular upgrades

Produce executive‑ready architecture options and recommendations for leadership decisions (build/partner/standardize)

Chiplet Partition, Resource Scaling, and System Topology

Drive functional partitioning across chiplets (compute, networking/I/O, accelerators, memory controllers, security/management), balancing PPA, D2D bandwidth/latency, validation complexity, and product flexibility

Define scalable system resource models (e.g., memory channels, PCIe lanes, pipeline scaling) and how these scale with chiplet count and topology

Specify any required logic / gaskets and integration constraints to enable modular assembly and consistent SW-visible behavior

Die‑to‑Die (D2D) Interconnect Architecture and QoS

Architect D2D communication for high bandwidth, low latency, and reliability across chiplets; define link budgets and requirements for bandwidth, latency, error handling, and flow control

Define inter-chiplet QoS mechanisms (e.g., arbitration, prioritization, isolation) and ensure the architecture supports workload-driven traffic patterns at scale

Align D2D choices with industry standardization direction (where applicable) and ensure project‑specific needs can be encapsulated cleanly

Chiplet System

Infrastructure: Power, Clock/Reset, Boot, telemetry

Define coordinated power delivery and power management flows across chiplets (telemetry, quiescence, package states, throttling), including system-level sequencing and corner cases

Architect clocking and reset distribution (reference clock delivery, local PLL strategies, reset sequencing, debug/manufacturing hooks) and ensure robust bring‑up across all dies

Drive the chiplet‑aware boot and early firmware architecture (e.g., parallel boot considerations, coordinated reset control) in partnership with FW/platform teams

RAS, Debug, and Observability Across Chiplets

Define cross‑chiplet error reporting, containment, and recovery policies, including consistent crashdump, telemetry access, and actionable observability for post‑silicon debug

Specify debug/trace infrastructure assumptions to ensure chiplet partitioning does not compromise lab efficiency or field diagnosability

Quantitative Trade Off Studies

Lead…