L

Lumilens

Photonic interconnects for AI infrastructure

RTL Design Lead/IC Front-End Architect/Datapath Lead

Silicon & ASIC DesignConfirmed live in the last 24 hours
Full-timeSan Jose, CA, USANo salary listedIn-person

About the Job

Requirements

  • 10+ years of digital design and micro-architecture experience, including lead ownership of a high-throughput datapath or SerDes-adjacent system-on-chip.
  • Experience designing the datapath of a network switch, network interface controller card, or memory controller at a major XPU or fabric vendor, or comparable high-throughput, high-reliability datapath experience.
  • Understanding of scale-up versus scale-out interconnect architectures and their bandwidth and latency trade-offs.
  • Strong SystemVerilog register-transfer-level design skills and timing-aware micro-architecture experience at multi-hundred-Gbps aggregate rates, with clock-domain-crossing discipline.
  • Working knowledge of UCIe, PCIe, and CXL datapath concepts and high-speed lane alignment and deskew.
  • Ownership of front-end design methodology, including lint and CDC/RDC signoff, clocking and reset architecture, design-for-test, and physical-closure collaboration.
  • Experience translating system requirements into an implementable, verifiable architecture specification.

Responsibilities

  • Own end-to-end verification for a first-of-its-kind chiplet and define the verification methodology, coverage strategy, and regression infrastructure before RTL implementation.
  • Define the digital front-end micro-architecture and architecture specification implemented by the RTL team.
  • Own front-end micro-architecture, block partitioning, RTL coding standards, lint, CDC/RDC signoff, and clean-RTL policy.
  • Own the high-speed datapath, including per-lane alignment and deskew, multiplexing, the UCIe streaming/RDI adapter, elastic FIFOs, and custom logic.
  • Own transport and link framing, mapping, and adaptation logic.
  • Develop timing- and floorplan-aware micro-architecture, including pipelining, retiming, and partitioning for physical closure at multi-hundred-Gbps rates with the backend team.
  • Define clocking, reset, and low-power architecture, including clock and reset domains, gating strategy, and UPF power intent.
  • Establish designer-level verification through block-level testbenches, sanity checks, and SystemVerilog Assertions coverage before handoff to design verification.
  • Provide architectural design-for-test and debug provisioning, including scan-friendly structure and on-die observability and controllability, with the design-for-test and design-verification leads.
  • Own IP integration, interface and register conventions, and parameterization for configurable lane counts and configurations.
  • Make architectural trade-offs with the modeling and design-verification leads regarding latency, buffering, datapath sampling, observability, verifiability, and debug.
  • Lead, grow, and mentor the front-end design team as the group scales.

Desired Qualifications

  • Experience with UCIe, PCIe, CXL, or CHI interface architecture and die-to-die or chiplet design.
  • Familiarity with forward error correction and coding datapaths, including Reed-Solomon, and gray-code mapping.
  • Background in optical transceivers, co-packaged optics, or linear-drive optics.
  • Experience with advanced-node N5, N3, or N2 design.

About the Company

Lumilens

Photonic interconnects for AI infrastructure

Lumilens designs and manufactures optical interconnect products to accelerate AI infrastructure. It provides high-density photonic interconnects—such as high-density pluggable, near-packaged, and co-packaged optics—built with in-house silicon photonics to scale GPU clusters in AI data centers. Its products include 800G and 1.6T optical transceivers for AI networking, and it is pursuing broader integration like Electrical-Optical Interposers through a strategic partnership with POET Technologies. The company differentiates itself through vertical integration, custom robotics-driven manufacturing, and AI-optimized production processes that aim to improve yield and supply chain stability. Lumilens targets hyperscalers and large AI providers, seeking to move interconnect bottlenecks from compute to scalable, efficient data-center networks. Its goal is to enable faster, denser GPU connectivity to support growing AI workloads and models.

Company size
51-200
Company stage
Series C
Total funding
$839.5M
Headquarters
San Jose, CA
Founded
2024

Recently Posted Jobs

Explore more opportunities at Lumilens

See more jobs