← All internships

Intern, LLM Inference Acceleration Architecture (Distributed Memory-Centric Accelerators)

Samsungsemiconductor

San Jose · Posted Nov 19

Apply

At a glance

Location
San Jose, CA
Work type
Onsite
Visa sponsorship
24 H-1B filings by this company
Category
Data

Internship in LLM inference acceleration architecture focused on distributed memory-centric accelerators. Develop and evaluate design-space exploration and architecture proposals for merge/reduction architectures, interconnect topologies, and precision of partial results for modern attention workloads.

What they are looking for

Architecture-level performance modelingDistributed memory ai acceleratorsGqa attentionMla attentionSparse top-k attentionHardware–software co-design

Get new Data internships by email

Similar roles

More at Samsungsemiconductor

Browse all internships More Data internships