Research Engineer Intern - AI Systems
Yotta Labs · Remote · United States · Hong Kong · Canada · Singapore ·
COMPENSATION
€•• – ••k
Listed on the employer's job board (Ashby)
- Fully remote (Global)
- Internship (12-16 weeks)
- CUDA + Triton + ROCm
- LLM systems optimization
- Return offer potential
Join Yotta Labs as a Research Engineer Intern to work on the next generation of AI cloud platforms. You will focus on optimizing compute kernels and runtime performance for NVIDIA, AMD, and AWS Trainium hardware.
This 12–16 week internship offers the chance to own a scoped project from design to production. Successful interns may receive return offers for full-time positions.
What you'll do
- Implement and optimize compute kernels for Attention, GEMM, MoE, and quantization.
- Build custom operators using CUDA, Triton, ROCm/HIP, or Neuron SDK with PyTorch/XLA.
- Profile and improve inference performance in vLLM, SGLang, and custom runtimes.
- Optimize kernel fusion, scheduling, KV-cache, and memory usage.
- Build benchmarks and resolve performance regressions using profiler traces.
- Ship code upstream to open-source AI infrastructure projects with tests.
What you'll need
- Pursuing BS, MS, or PhD in Computer Science, Engineering, or related field.
- Solid programming skills in Python and familiarity with C++.
- Understanding of GPU/accelerator architecture fundamentals like memory hierarchy.
- Experience writing CUDA, Triton, ROCm/HIP, or Neuron kernels.
- Knowledge of AI frameworks such as PyTorch, Dynamo, or LMCache.
- Familiarity with profiling tools like Nsight, ROCm Profiler, or Neuron Profiler.
- Ability to work independently in a collaborative remote environment.
Nice to have
- Contributions to open-source AI infra projects like vLLM, SGLang, or PyTorch.
- Familiarity with LLM inference internals such as FlashAttention or PagedAttention.
- Experience with performance debugging on real workloads.
- Publications in top-tier conferences like MLSys, OSDI, or SC.
Stack
- Python
- C++
- CUDA
- Triton
- ROCm/HIP
- Neuron SDK
- PyTorch
- XLA
- vLLM
- SGLang
How you'll work
- Fully remote position available globally.
- Internship duration of 12 to 16 weeks.
- Flexible remote work environment.
What you get
- Competitive internship compensation.
- Direct mentorship from engineers at leading institutions.
- Access to latest-generation NVIDIA GPUs, AMD accelerators, and AWS Trainium.
- Fast path to full-time return offer for top performers.
Hiring process
- Apply directly or send resume via email.
- Include links to relevant projects, GitHub, or open-source PRs.
- Project contributions are valued more than cover letters.