About the role#
The Seed Infrastructures team manages distributed training, reinforcement learning frameworks, high-performance inference, and hardware compilation for AI foundation models. This internship provides PhD students the opportunity to contribute to research and product development while working alongside industry experts.
What you'll need#
- Currently pursuing a PhD in Computer Science, Electrical Engineering, or a related technical field.
- Strong programming skills in Python or C++.
- Solid understanding of systems, distributed computing, machine learning systems, or performance optimization.
- Experience with distributed training frameworks like PyTorch FSDP or Megatron-style parallelism.
- Experience with reinforcement learning training systems.
- Experience with GPU programming such as CUDA or Triton, or compiler technologies.
- Preferred: Experience with large-scale ML systems, contributions to open-source ML tools, or publications in distributed systems and ML fields.
What you'll do#
- Design and optimize large-scale distributed training systems, focusing on parallelism, memory efficiency, and fault tolerance.
- Contribute to reinforcement learning training frameworks and post-training systems.
- Enhance inference performance, including latency and throughput for foundation models.
- Build compiler or runtime optimizations for GPUs and other accelerators.
- Conduct system-level performance analysis, profiling, and bottleneck diagnosis.
- Create tools and automation to increase developer productivity and system reliability.
Location & details#
- Location: San Jose, California.
- Modality: On-site.
- Employment: Full-time internship.
- Compensation: $85 per hour.
- Review process: Rolling basis.


