Research Engineer Intern – AI Systems

This is a fully remote position, open to applicants in United States, +3 more countries.

📋 Description

• Develop and enhance compute kernels for Attention, GEMM, MoE, and quantization on platforms such as NVIDIA, AMD, or AWS Trainium.

• Create custom operators utilizing CUDA, Triton, ROCm/HIP, or the Neuron SDK in conjunction with PyTorch/XLA.

• Analyze and optimize inference performance in vLLM, SGLang, and our proprietary runtimes — focusing on kernel fusion, scheduling, KV-cache, and memory enhancements.

• Construct benchmarks, identify performance regressions, and convert profiler traces into tangible performance improvements.

• Contribute code to open-source AI infrastructure initiatives, ensuring proper testing and documentation.

⛳️ Requirements

• Currently enrolled in a BS, MS, or PhD program in Computer Science, Computer Engineering, or a related discipline.

• Proficient programming skills in Python and a working knowledge of C++.

• Familiarity with the fundamentals of GPU/accelerator architecture (memory hierarchy, parallelism, occupancy) obtained through coursework, research, or projects.

• Experience in writing CUDA, Triton, ROCm/HIP, or Neuron kernels — including academic and personal projects.

• In-depth understanding of AI frameworks (such as PyTorch, Dynamo, LMCache), model architectures, and profiling tools (including Nsight, ROCm Profiler, or Neuron Profiler).

• Strong analytical and problem-solving abilities, with the capacity to work autonomously in a collaborative remote setting.

🏝️ Benefits

• Competitive internship compensation

• Flexible remote work environment

• Direct mentorship from engineers from leading institutions and tech companies

• Fast path to a full-time return offer for top performers