AI Accelerator Architect

AI Accelerator Architect

Full-Time On-site
E

We are seeking an experienced Principal ASIC Design Engineer to contribute to the architecture, microarchitecture, and RTL development of next-generation AI accelerators and high-performance SoC subsystems.

The ideal candidate will have extensive hands-on design experience in GPU, NPU, or complex SoC subsystems, with a proven track record of delivering advanced ASIC products from initial specification through RTL implementation, verification, physical design, silicon bring-up, and production.

This position focuses on hands-on engineering, complex digital design, performance optimization, and implementation of advanced computing and interconnect architectures.

Responsibilities

1. AI Accelerator and SoC Architecture

  • Develop architecture and microarchitecture for high-performance AI accelerators and complex SoC subsystems.
  • Translate system-level requirements into detailed functional and microarchitecture specifications.
  • Design compute pipelines, memory hierarchies, on-chip interconnects, DMA engines, and data movement architectures.
  • Evaluate design trade-offs involving performance, power, area (PPA), memory bandwidth, latency, and scalability.
  • Collaborate with system architects, algorithm engineers, and software engineers to optimize hardware implementations for AI/ML workloads.

2. Microarchitecture and RTL Design

  • Develop detailed microarchitecture specifications and synthesizable RTL implementations using Verilog/SystemVerilog.
  • Design complex digital datapaths, control logic, finite state machines, arbitration mechanisms, and communication interfaces.
  • Implement efficient pipelining, buffering, flow control, and resource-sharing mechanisms.
  • Optimize RTL for timing, power, area, and implementation feasibility.
  • Perform RTL simulation, debugging, lint, clock-domain crossing (CDC) analysis, and synthesis-related design checks.
  • Work with verification and physical design engineers to resolve functional, timing, and implementation issues.

3. High-Speed Interfaces and Memory Subsystems

  • Design and integrate high-bandwidth memory and high-speed I/O subsystems.
  • Work with memory interfaces and controllers, including LPDDR5/LPDDR5X and LPDDR6.
  • Design or integrate PCIe-based subsystems, including PCIe Gen5/Gen6 controllers, DMA engines, and associated data movement infrastructure.
  • Understand NoC architectures, memory hierarchies, cache coherency, and bandwidth optimization.
  • Optimize data movement and memory access efficiency for high-performance computing and AI accelerator applications.

4. ASIC Development from Specification to Production

  • Participate in the complete ASIC development lifecycle, from specification and microarchitecture through RTL design, verification, synthesis, physical implementation, tape-out, and silicon validation.
  • Develop production-quality RTL that meets functional, performance, power, and area requirements.
  • Collaborate with verification engineers on block-level and SoC-level verification plans, test cases, and functional coverage.
  • Support lint, CDC/RDC analysis, formal verification, synthesis, static timing analysis, and design-for-test activities.
  • Debug functional and timing issues identified during simulation, emulation, FPGA prototyping, and physical implementation.
  • Support silicon bring-up, post-silicon debugging, performance characterization, and production validation.

5. Hardware/Software Co-Design and Verification

  • Collaborate with firmware, driver, compiler, and system software engineers to define and implement hardware/software interfaces.
  • Participate in hardware/software partitioning and optimization for AI accelerator applications.
  • Develop register interfaces, control/status mechanisms, interrupt handling, and system-level data flows.
  • Support IP-level and SoC-level functional verification using simulation, emulation, and FPGA prototyping.
  • Debug hardware/software integration issues and contribute to system-level validation.

Qualifications

  • Bachelor's or Master's degree in Electrical Engineering, Computer Engineering, Computer Science, or a related field.
  • 10+ years of hands-on experience in digital ASIC/SoC microarchitecture and RTL design.
  • Demonstrated design experience in at least one of the following: GPU or parallel computing architectures; NPU, AI accelerator, or machine learning processor architectures; Complex SoC subsystems, such as memory controllers, interconnect fabrics, DMA engines, or high-performance processing subsystems
  • Proven experience developing ASIC/SoC designs from specification through tape-out and silicon production.
  • Strong expertise in microarchitecture development and RTL design using Verilog/SystemVerilog.
  • Solid understanding of digital datapaths, control logic, pipelining, memory subsystems, on-chip interconnects, and clock/reset architectures.
  • Working knowledge of PCIe and LPDDR memory technologies, including controller architecture, integration, and system-level considerations.
  • Experience in at least one of the following: Hardware/software co-design and system integration; IP-level or SoC-level functional verification
  • Familiarity with ASIC design methodologies and tools, including RTL simulation, synthesis, static timing analysis, lint, and CDC.
  • Strong debugging, analytical, and problem-solving skills.
  • Ability to independently develop, implement, and debug complex ASIC functional blocks and subsystems.

Preferred Qualifications

  • Direct experience designing GPU, NPU, AI accelerator, or heterogeneous computing SoCs.
  • Experience with chiplet architectures and die-to-die interconnects.
  • Experience with PCIe Gen5/Gen6 and advanced memory technologies such as LPDDR6.
  • Knowledge of NoC architectures, cache coherency, memory consistency, and multi-die system architectures.
  • Experience with high-bandwidth data movement, DMA engines, and accelerator memory optimization.
  • Understanding of AI/ML workloads, including matrix multiplication, tensor operations, transformer models, and inference acceleration.
  • Familiarity with AMBA AXI/ACE/CHI or other industry-standard SoC interconnect protocols.
  • Experience with UVM-based verification, formal verification, FPGA prototyping, or hardware emulation.
  • Familiarity with advanced semiconductor process technologies, including 7nm, 5nm, 3nm, or below.
  • Experience with low-power design techniques, including clock gating, power domains, and power optimization.

#J-18808-Ljbffr

E

Contact Details:

Ecarx Co. Recruitment Team