LLM Inference Acceleration Architecture Intern - Distributed Memory-Centric Accelerators
SamsungAdded 12h ago
San Jose, CAInternship
Save this job and browse the whole board.
Create free account
No credit card needed
What you'll do
- Model and simulate LLM attention workloads to optimize distributed memory-centric AI accelerators.
- Analyze merge and reduction bottlenecks in modern attention architectures like GQA and MLA.
- Design hardware strategies for data placement and token selection to minimize communication overhead.
- Present research findings and technical reports to the architecture team at summer's end.
What they're looking for
- Current Master’s or PhD student in engineering or computer science.
- Proficiency in Python or C++ with experience in architecture simulators.
- Strong knowledge of Transformer inference, memory systems, and interconnects.
Pay and perks
- Hourly base pay ranges from $28 to $68 USD.
- Includes relocation and housing stipends for eligible interns.
Summarized by newgrad.ai from Samsung's public posting. Check the original for full details before you apply.
Apply on Samsung's site.
Apply nowListing wrong or expired, or you're the employer and want it removed? Let us know.
