2 months ago
Palo Alto, CA, USAIntern
H1B sponsor
Responsibilities
- Contribute to development of an ML compiler that maps autonomy machine learning models to the Rivian Autonomy Processor (RAP1).
- Design and implement hardware-aware optimizations including quantization, model compression, memory-efficient representations, and operator fusion.
- Collaborate with hardware teams to co-optimize model architectures and compute pipelines for latency, throughput, and power constraints.
- Benchmark and analyze system performance across platforms and iterate toward efficient deployment.
- Partner with autonomy teams to align model optimization with the hardware roadmap and real-world autonomy requirements.
Requirements
- Currently pursuing a master’s or PhD degree at an accredited university in Computer Engineering or a closely related field.
- Excellent C/C++ and Python programming skills.
- Experience with system-on-chip platforms used for machine learning.
- Strong understanding of deep learning software models.
- Experience with compiler pipeline development is preferred.
- Proficiency with deep learning frameworks and their low-level intermediate representations or export formats.
Benefits
- Spring 2027 co-op runs from January through August 2027 and requires onsite, full-time work at 40 hours per week for the entire term.
- Benefits include paid vacation, paid sick leave, and medical insurance benefits.
- Internship positions are not eligible for retirement benefits.
- Applications are accepted on an ongoing basis until the role is filled or the opening is no longer needed.
Categories
About Rivian
Rivian designs and manufactures electric vehicles, including the R1T pickup, R1S SUV, and commercial delivery vans, selling directly to consumers and fleet operators. Founded in 2009 and headquartered in Irvine, California, it operates a manufacturing plant in Normal, Illinois, and is publicly traded on NASDAQ. The company also develops in-vehicle software and electrical architecture and offers charging and service support for its owners.
