Platform Engineering team Bing Platform Engineering
Microsoft's Bing Platform Engineering team is seeking highly motivated, independent hardware design engineers at all career stages to join us in blazing the trail for FPGA-based acceleration at datacenter scale. By leveraging the huge amounts of fine-grained parallelism delivered by current and future FPGAs, reconfigurable computing can radically accelerate many types of computations. We are applying this technology to challenging applications important to Microsoft and our customers, including but not limited to Bing search. Our work includes developing, optimizing, deploying, and maintaining FPGA accelerators, infrastructure, and tools. Our current focus is on accelerating deep neural networks (DNNs) via the Brainwave accelerator architecture. We are heavily involved in developing, extending, and deploying Brainwave, and in bringing DNN models to production on FPGAs using Brainwave. Working with Microsoft model developers to deploy their models on Brainwave gives us the unusual opportunity
Primary responsibilities include: Develop, maintain, and extend high-performance FPGA-based computing applications. Deploy and maintain systems to meet significant reliability and service level requirements. Work in a team of hardware and software engineers to deploy working systems to a very large user base. Collaborate closely with other product groups, including Office and Azure.
A solid computer engineering background (MS or PhD in computer engineering or related field, or equivalent experience) Experience with FPGA application RTL design (Verilog, SystemVerilog, or VHDL) and FPGA-based systems. Previous software or hardware development experience in all stages of a release cycle (design, implement, and test) with ability to cover multiple job roles. Experience working in a team environment. Experience in system-level design, especially in high-performance, distributed and/or fault-tolerant systems. Software development experience, particularly high-performance C/C++ system