Role Overview
NVIDIA seeks an AI Compiler and Kernel Intern to apply AI agents, machine learning, and performance engineering to compilers, GPU kernels, and developer tools. The intern will collaborate with compiler, kernel, and AI engineers on a scoped project involving code, compiler and kernel feedback, performance experiments, and optimization. Focus areas may include agentic workflows, learned models, compiler optimization, GPU kernel development, performance analysis, or AI-enabled developer tools.
Responsibilities
- Build AI-agent workflows interacting with code, compilers, GPU kernels, benchmarks, performance tools, and engineering systems.
- Develop tools for agents to collect data, run experiments, analyze feedback, identify optimization opportunities, and recommend next steps.
- Explore data-driven approaches for compiler optimization, GPU kernel generation/tuning, performance prediction, decision support, and workflow automation.
- Build reliable interfaces between AI agents and compiler, kernel, or developer-tool infrastructure.
- Prototype improvements to compiler flows, code generation, GPU kernels, optimization tools, experimentation systems, or developer workflows.
- Collaborate with compiler, kernel, architecture, AI, and performance engineers to evaluate ideas on representative workloads.
- Measure correctness, runtime performance, compile time, resource utilization, reliability, and other outcomes.
- Document design, experiments, results, limitations, and next steps.
Requirements
- Pursuing BS, MS, or PhD in Computer Science, Computer Engineering, Electrical Engineering, or similar.
- Strong programming skills in Python, C, C++, or similar.
- Solid foundations in data structures, algorithms, software engineering, and quantitative problem solving.
- Coursework, research, or project experience in AI, machine learning, AI agents, compilers, programming languages, GPU kernels, CUDA/accelerator programming, systems, performance engineering, optimization, or developer tools.
- Ability to design experiments, analyze results, investigate technical problems, and communicate findings clearly.
- Experience in one or more of AI, compiler, kernel, performance, or systems areas, with ability to learn across technical boundaries.
Nice to Have
- Experience building AI agents, LLM applications, evaluation systems, coding workflows, or developer tools.
- Experience with LLVM, MLIR, compiler optimization passes, code generation, domain-specific languages, or compiler infrastructure.
- Experience writing, analyzing, or optimizing GPU kernels using CUDA, Triton, or similar accelerator programming model.
- Hands-on experience with reinforcement learning, evolutionary algorithms, predictive modeling, ranking, search, or high-dimensional data analysis.
- Research, internships, open-source contributions, or personal projects measuring runtime, compile time, resource use, code quality, or developer productivity.