Software Engineer Project Intern (Model Infrastructure) - 2026 Start (BS/MS)
Core
Build ultra-high-performance infrastructure for recommendation models, optimizing training and inference pipelines for systems with hundreds of billions of parameters.
Role type
Software Engineering Intern (Model Infrastructure)
Builds
High-concurrency engines for Petabyte-scale streaming training and specialized systems for LLM integration into recommendation stacks.
Domain
AI Infrastructure / Recommendation Systems / High-Performance Computing
Deliverable
production ML models
Required skills
C++, Python, CUDA, PyTorch, Computer Architecture, Distributed Systems
Preferred skills
Transformer architectures, 3D parallelism, Torch.compile stack, High-performance kernels, RDMA networking
Technologies
CUDA, Triton, NCCL, PyTorch, TensorFlow, RDMA, NVMe, HBM, DDR
Responsibilities
Optimize training and inference pipelines to maximize hardware utilization; Architect systems for LLM integration focusing on memory-efficient attention and KV cache; Build engines for Petabyte-scale streaming training; Design recommendation architectures optimized for GPU/NPU interconnects; Innovate on storing and synchronizing massive model states across heterogeneous memory hierarchies.