Software Engineer Intern, AI and DL Kernel Libraries - 2027
Core
Develop foundational software and low-level kernels for NVIDIA's AI platform, focusing on deep learning primitives, runtime systems, and performance-critical GPU workloads.
Role type
Software Engineer Intern (AI/DL Kernel Libraries)
Builds
NVIDIA's AI software stack including cuDNN, FlashInfer, and optimized support for large language model inference workloads.
Domain
AI / Deep Learning / GPU Systems / High-Performance Computing
Deliverable
production ML models
Required skills
C/C++, Python, CUDA, Deep Learning frameworks (PyTorch, JAX, TensorFlow, ONNX), Linear Algebra, Performance Analysis, Code Optimization
Preferred skills
Inference engines (vLLM, SGLang, TensorRT-LLM), Domain-specific compilers (MLIR, Apache TVM), GPU performance modeling, Open-source contributions
Technologies
CUDA, cuDNN, FlashInfer, PyTorch, JAX, TensorFlow, ONNX, vLLM, SGLang, TensorRT-LLM, MLIR, Apache TVM, TensorIR
Responsibilities
Design and optimize kernels for high-impact AI workloads; Build extensible software abstractions for deep learning libraries; Contribute to just-in-time compilation and code generation systems; Analyze workload performance and tune software; Collaborate with teams on deep learning frameworks, compilers, and GPU architecture.
Seniority
Intern