Physical Design Methodology Engineer, Annapurna Labs

Amazon Web Services (AWS) is seeking an experienced Physical Design Engineer to join the Cloud-Scale Machine Learning Acceleration team. This role focuses on designing and optimizing hardware for AWS data centers, particularly for the AWS Inferentia machine learning inference product. The position demands high standards and a commitment to improving product performance, quality, and cost in a fast-paced environment.

Responsibilities:

• Create and support innovative physical design methodology and CAD flows.
• Develop cloud infrastructure to support physical design work.
• Drive improvement in RTL2GDS flows/methodology for PPA and TAT improvement.
• Create dashboard/central reports for project tracking and visualizing QoR/stats.
• Interface with RTL, Physical Design, Package Design, DFT, and other teams to enhance methodologies and efficiencies.
• Collaborate with EDA tool vendors to evaluate new tools, resolve bugs, and improve usability.

Qualifications:

• Bachelor's or Master's degree in Electrical Engineering, Computer Engineering, or Computer Science.
• Minimum of 3+ years in developing design methodology or CAD flows in synthesis, PNR, or sign-off areas for advanced technology nodes.
• Experience in writing production scripts for implementation and sign-off tools in TCL, Perl, and/or Python.
• Solid understanding of ASIC physical design, flows, and methodologies including synthesis, place and route, STA, and formal verification.
• Proven track record of delivering metric-driven PPA flow development and support.

Skills:

• Expertise in PD tools such as Innovus, ICC2, FusionCompiler, STA, and Sign-Off.
• Experience in evaluating multiple vendor solutions and driving tool decisions.
• Knowledge of high-performance, low-power physical design and implementation techniques with industry-standard tools.
• Excellent programming skills in Python, Perl, TCL, Shell, etc.
• Good understanding of algorithms with an emphasis on optimization.
• Experience with machine learning.