About Altera

At Altera™, our independence as the world’s largest pure‑play FPGA solutions provider gives us the focus, speed, and agility to innovate without compromise. With more than four decades of industry‑leading FPGA expertise, our singular mission is to deliver the programmable technologies that help customers differentiate, innovate, and scale across rapidly evolving markets like AI, cloud, networking, and edge. As an independent company, we move faster, invest deeper, and partner more closely—empowering our teams to drive breakthrough innovation and shape the future of the FPGA industry.

About the Role

Altera is seeking a highly experienced Clocking Architect to lead the definition, design, and integration of clocking architectures for its next-generation FPGA and SoC devices. In this role you will own the end-to-end clocking strategy — from subsystem clock planning through full-chip integration — ensuring robust clock delivery, minimal skew, and compliance with all functional and DFT timing requirements across a diverse set of high-speed protocols.

The ideal candidate brings deep hands-on expertise in PCIe, High-Speed Memory I/O, ARM, Ethernet, and SerDes clocking domains, combined with mastery of SDC constraint authoring for both functional and DFT modes. This is a high-visibility, high-impact role with direct influence over silicon architecture decisions.

Clocking Architecture & Full-Chip Strategy

• Own and drive the complete clocking architecture for Altera’s FPGA/SoC devices, including clock tree topology, domain partitioning, frequency planning, and PLL/DLL resource allocation.

• Define subsystem-level clocking plans aligned with chip-level power budgets, protocol timing margins, and physical implementation constraints.

• Establish clock gating policies, low-power clocking methodologies, and dynamic frequency scaling strategies.

• Deliver the clocking specification, clock architecture diagrams, and constraint management documentation as program-level deliverables.

• Collaborate with RTL, physical design, timing, and verification teams to ensure the clock architecture is correctly modeled and implemented across all design stages.

Protocol-Specific Clocking Expertise

• PCIe (Gen4/5/6): Define REFCLK, PCLK, and spread-spectrum clocking (SSC) architectures; support SRIS/SRNS topologies and PCIe reset/power management clocking sequences.

• High-Speed Memory I/O (DDR5, LPDDR5, HBM2E/HBM3): Architect DFI clock, write leveling, read path, and DQS clocking to meet memory controller timing requirements.

• ARM Subsystems (Cortex-A/M/R, DSU, CMN Interconnect): Own core, peripheral, and AMBA bus clock hierarchies; ensure compliance with ARM’s timing and power requirements.

• Configuration Interfaces (eSPI, SPI, JTAG, LPC): Define boot and configuration clock domains, including secure boot and JTAG test clock isolation.

• Ethernet (1G/10G/25G/100G/400G): Architect MAC, PCS, and PHY clock domains; support SyncE and IEEE 1588 Precision Time Protocol (PTP) clocking requirements.

• SerDes (PCIe, Ethernet, Custom High-Speed I/O): Define lane CDR, TX/RX PLL, and gearbox clocking; address reference clock distribution and jitter budgets for multi-lane interfaces.

ML/AI Accelerator Clocking

• Architect clocking solutions for FPGA-based and SoC-integrated ML/AI inference and training accelerator pipelines, including matrix multiply units, systolic arrays, and vector processing engines.

• Define multi-frequency clock domain strategies for AI workloads: separate clock planes for compute fabric, memory subsystem (HBM/LPDDR), host interconnect (PCIe), and control path to maximize throughput and efficiency.

• Own clocking for high-bandwidth AI memory interfaces (HBM2E/HBM3, LPDDR5X) used in AI inference cards and data-center FPGA overlays, ensuring sub-picosecond jitter budgets at peak bandwidth.

• Support dynamic voltage and frequency scaling (DVFS) clocking architectures that adapt compute clock rates to AI workload demand and thermal headroom.

• Collaborate with AI IP and compiler teams to align clock domain boundaries with dataflow partitioning, pipeline stage timing, and latency requirements of ML models.

• Define clocking for