About Rivian Rivian is on a mission to keep the world adventurous forever. This goes for the emissions-free Electric Adventure Vehicles we build, and the curious, courageous souls we seek to attract. As a company, we constantly challenge what’s possible, never simply accepting what has always been done. We reframe old problems, seek new solutions and operate comfortably in areas that are unknown. Our backgrounds are diverse, but our team shares a love of the outdoors and a desire to protect it for future generations. Role Summary As a Staff Software Engineer, ML training and inference infrastructure, you will be a member of the Perception team at Rivian, which develops advanced machine learning algorithms that directly impact safety critical self-driving features of our category defining vehicles. We are looking for candidates with deep knowledge and strong enthusiasm towards establishing a state-of-art ML infrastructure for training and inference of large autonomous driving models; and optimizing the training and inference performance. Responsibilities Optimize the performance of Deep Learning training workload on NVIDIA GPU systems on a large scale Optimize the latency of model inference and model pre- and post-processing on onboard systems Design, train, and deploy large deep learning models that can leverage the vast amount of labeled and unlabeled data Qualifications PhD in CS/CE/EE, or equivalent, in industry experience Deep knowledge of PyTorch Knowledge of model training framework (e.g. PyTorch Lightning, ray, etc.) In-depth knowledge of transformer architecture and ways to accelerate the training and inference of transformer models Experience of performing large scale distributed training of models A track record of profiling models and doing detective work to improve model training and inference speed Preferred Skill Requirements: Experience with CUDA or Triton language for writing custom ops Knowledge of Nvidia TensorRT Knowledge of NCCL Experience with edge computing systems A track record of efficiently solving complex problems collaboratively on larger teams Pay Disclosure Salary Range for California Based Applicants: $228,000.00 - $285,000.00 (actual compensation will be determined based on experience, location, and other factors permitted by law). Benefits Summary: Rivian provides robust medical/Rx, dental and vision insurance packages for full-time employees, their spouse or domestic partner, and children up to age 26. Coverage is effective on the first day of employment Equal Opportunity Rivian is an equal opportunity employer and complies with all applicable federal, state, and local fair employment practices laws. All qualified applicants will receive consideration for employment without regard to race, color, religion, national origin, ancestry, sex, sexual orientation, gender, gender expression, gender identity, genetic information or characteristics, physical or mental disability, marital/domestic partner status, age, military/veteran status, medical condition, or any other characteristic protected by law. Rivian is committed to ensuring that our hiring process is accessible for persons with disabilities. If you have a disability or limitation, such as those covered by the Americans with Disabilities Act, that requires accommodations to assist you in the search and application process, please email us at candidateaccommodations@rivian.com. Candidate Data Privacy Rivian may collect, use and disclose your personal information or personal data (within the meaning of the applicable data protection laws) when you apply for employment and/or participate in our recruitment processes (“Candidate Personal Data”). This data includes contact, demographic, communications, educational, professional, employment, social media/website, network/device, recruiting system usage/interaction, security and preference information. Rivian may use your Candidate Personal Data for the purposes of (i) tracking interactions with our recruiting system; (ii) carrying out, analyzing and improving our application and recruitment process, including assessing you and your application and conducting employment, background and reference checks; (iii) establishing an employment relationship or entering into an employment contract with you; (iv) complying with our legal, regulatory and corporate governance obligations; (v) recordkeeping; (vi) ensuring network and information security and preventing fraud; and (vii) as otherwise required or permitted by applicable law. Rivian may share your Candidate Personal Data with (i) internal personnel who have a need to know such information in order to perform their duties, including individuals on our People Team, Finance, Legal, and the team(s) with the position(s) for which you are applying; (ii) Rivian affiliates; and (iii) Rivian’s service providers, including providers of background checks, staffing services, and cloud services. Rivian may transfer or store internationally your Candidate Personal Data, including to or in the United States, Canada, the United Kingdom, and the European Union and in the cloud, and this data may be subject to the laws and accessible to the courts, law enforcement and national security authorities of such jurisdictions. Please note that we are currently not accepting applications from third party application services.
Staff Software Engineer, ML Training and Inference Infrastructure
Staff Software Engineer focused on optimizing ML training and inference infrastructure for large autonomous driving models on NVIDIA GPU systems, including on-board inference latency optimization and distributed training.
What you'd actually do
- Optimize the performance of Deep Learning training workload on NVIDIA GPU systems on a large scale
- Optimize the latency of model inference and model pre- and post-processing on onboard systems
- Design, train, and deploy large deep learning models that can leverage the vast amount of labeled and unlabeled data
Skills
Required
- PyTorch
- model training frameworks
- transformer architecture
- large scale distributed training
- model profiling
- inference optimization
Nice to have
- CUDA
- Triton
- Nvidia TensorRT
- NCCL
- edge computing systems
What the JD emphasized
- Deep knowledge of PyTorch
- Knowledge of model training framework (e.g. PyTorch Lightning, ray, etc.)
- In-depth knowledge of transformer architecture and ways to accelerate the training and inference of transformer models
- Experience of performing large scale distributed training of models
- A track record of profiling models and doing detective work to improve model training and inference speed
Other signals
- optimize training performance
- optimize inference latency
- design, train, and deploy large deep learning models
- large scale distributed training
- transformer architecture acceleration