Student Researcher - (Seed Model - Infra - Compiler) - 2026 Start (PhD)
AI Compiler Optimization
Likely openWe list a role only while we believe it is still open. Confirm on the posting before you apply.
What stands out
- Pay is stated up front — $85 / hour. Most postings never say.
- Posted in the last 24 hours.
The gist
This PhD-level role involves optimizing AI compilers for training and inference, with a focus on GPU and NPU performance. The intern will work on compiler passes, model execution, and deployment pipelines, collaborating closely with researchers and engineers. Ideal for students with a strong background in ML frameworks and compiler technologies.
What you would actually do
- Optimize model execution on GPU and NPU accelerators
- Develop MLIR-based compiler passes for graph lowering
- Collaborate with researchers on compiler and runtime improvements
Skills mentioned
About this listing
- Found on
- First seen by InternDoor
- , within 2 hours of being posted
- Last seen listed
- Application page
- joinbytedance.com
InternDoor has recorded 70 engineering postings from ByteDance in the US since Aug 2026. LinkedIn listings stay on InternDoor for at most 30 days after we first see them, so this one comes down by , or sooner if we find its application has closed. When it comes down, this address points to ByteDance’s page instead of a dead end.
How to apply
Internships in the US often collect hundreds of applicants within a day, so applying early matters more than applying perfectly. A half-finished application sent on the first morning beats a polished one sent on the third.
This summary was written by InternDoor from the public posting, and is not the employer's own wording. The linked posting is the source of truth — check it before you apply.