SMTS Software Development Eng- ML Framework and Runtime

THE ROLE

AMD is looking for a talented, self-driven and motivated engineer to technically lead AIG s Vitis AI Compiler projects working on AMD s XDNA (AI Engine) architecture and the Vitis AI family of software tools. The XDNA is an industry leading NPU (Neural Processing Engine) architecture in terms of performance per watt and is used in AMD s client and embedded devices as the primary engine for Machine Learning workloads. It is the hardware engine behind Windows Co-pilot on AMD devices.

The team provides a fast-paced environment offering each of its members immense opportunity to interact with a wide variety of people including from other organizations like hardware designers, marketing, support, and even direct customer interaction, and truly learn and grow their skills and capabilities.

THE PERSON :

The ideal candidate should be passionate about software engineering and possess leadership skills to drive sophisticated technical issues to resolution. They should have demonstrated ability to identify technical problems, explore and propose viable options, and apply technical solutions. They should be able to excel in a global team environment with strong verbal and written communication skills.

KEY RESPONSIBILITIES :

• Vitis AI is AMD s primary SDK that enables users to compile and run their ML models on the XDNA architecture which forms the basis for AMD s. As a senior member of this high-performance team, the selected candidate will have the opportunity to work on integrating the ML tool chain into frameworks like ONNX, Pytorch, TensorFlow etc.
• Candidate will have opportunity to work on orchestrating the compilation of ML model through different phases
• Candidate will integrate runtime execution of ML model on the NPU hardware through the runtime and driver.
• Candidate will collaborate with compiler and runtime teams to bring up latest AI models like CNNs, Transformers, Stable Diffusion, NLPs etc. on the XDNA simulator.
• Candidates would develop a deeper understanding of the various ML models, and how they are executed, identify performance bottlenecks and enable faster development.

PREFERRED EXPERIENCE :

ACADEMIC CREDENTIALS :

Skills Required

C++, Pytorch, Github, Docker, Cmake