Deep Learning Compiler (TVM) Developer

East Palo Alto, California, USA

Full Time logo
Apply now Apply later

Posted 2 days ago

At AWS AI, we want to make it easy for our customers to deploy machine learning models in the cloud or at the edge. SageMaker provides a complete set of services to simplify the workflow of building, training, and deploying ML models. As one of the SageMaker services, Neo provides a deep learning compiler and runtime that is designed to run any machine learning model on any hardware.

Neo uses open source Apache TVM along with partner-provided compilers to optimizes machine learning models. After compilation, models perform at up to twice the speed of the original framework with no loss in accuracy. Upload a pre-trained model built with MXNet, TensorFlow, PyTorch, or XGBoost to your S3 bucket, choose your target hardware platform from Intel, NVIDIA, ARM, Ambarella, Qualcomm, NXP, or TI. With a single API call, SageMaker Neo optimizes the model, converts it into an executable module, and returns it to your S3 bucket. Then the free open source Neo runtime uses less than 100th of the space of the framework to run the model on the target hardware.

The SageMaker Neo team is growing rapidly to keep up with growth in customers and their requests. We are hiring well-rounded applied scientists and software developers with backgrounds in machine learning, compilers, systems, and AI accelerators. If you have worked on HPC and performance tuning, you will enjoy working on the breadth of ML applications that we optimize.

As a deep learning compiler developer, you will work on systematic approaches to improve the performance of deep learning models in Apache TVM. You will help develop the compilation service and runtime for machine learning while working at the frontier of the field where our team publishes its research. The work offers an extremely broad set of opportunities to work as a full stack SDE with exposure to multiple AI applications, ML frameworks, models, compilers, systems SW, and various AI hardware including ARM, Intel, Nvidia, AWS AI accelerators, and emerging edge AI ASICs.

Join the Amazon SageMaker Neo team to help AWS customers deploy machine learning models in the cloud and on edge devices at scale in production. Work on an open source industry-standard compiler and runtime for machine learning that is already deployed on millions of devices.

Basic Qualifications

· Master's Degree in Computer Science or Engineering

· 4+ years of software development experience in high performance computing, machine learning, systems architecture, or related areas

Preferred Qualifications

· Proven ability to develop and deliver an optimizing compiler for high level / domain specific programming language

· 6+ years technical leadership role in machine learning, HPC or related areas

· PhD in Computer Science

Amazon is an Equal Opportunity-Affirmative Action Employer – Minority / Female / Disability / Veteran / Gender Identity / Sexual Orientation

Job tags: AI AWS Deep Learning Engineering Machine Learning ML MXNet Open Source PyTorch Research TensorFlow XGBoost