Alibaba DAMO Academy-Near Data Processing Accelerator Architect-Sunnyvale
Develop domain specific near data processing architecture and system solution for data center/cloud computing, in memory database, and machine learning applications.
· Define system-level accelerator architectures including key features, performance requirements, block level definition.
· Develop FPGA/ASIC PoC architecture, design and feasibility studies for domain specific near data processing architecture.
· Conduct AI system performance study, workloads characterization.
· Propose and execute on innovations in large scale AI system software and hardware architecture. Help define new projects to address these opportunities and play a key technical role in their execution.
· Documentation of architecture, design and IP specifications
· Create innovative IP, publish at conferences, and generate whitepapers.
Ph.D. in Computer Architecture, Electrical Engineering, Computer Science or related field.
· >5 years of experience in AI system, memory and/or storage system architecture or design.
· Demonstrated expertise with FPGA design and synthesis tools, simulation tools, and scripting tools for test automation; In-depth knowledge in HDL/Verilog development, IP integration, debug.
· Good understanding of neural network accelerator architecture with GPU/FPGA etc., and deep learning infrastructure (Torch, Caffe, TensorFlow, etc.)
· Fundamental understanding of computer architecture, server memory and storage hierarchies, and the potential performance trade-offs and optimizations involved.
· Track record of innovation and creativity in problem solving.
· Highly motivated with excellent verbal and written communication skills.