Deep Learning Performance Software Intern - 2027
Originally posted 30 September 2026 by the employer.
About the role
What you'll do
- Create and maintain SKILL, Wiki, and agent harness.
- Develop TileGym, Triton CUDA TileIR backend, and CUDA Tile.
- Develop highly optimized deep learning kernels through tile-based GPU programming model.
- Conduct end-to-end performance optimization through tile-based GPU programming model.
- Perform performance optimization, analysis, and tuning.
What you'll need
- Pursuing a degree in an engineering or computer science related field.
- Understanding of agentic systems, including LLM APIs, prompting, tool use/function calling, agent loops, planning, reasoning, memory, RAG, skills, MCP, sub-agents, multi-agent architectures, context engineering, and harness engineering.
- Excellent C/C++ programming and software design skills.
- Performance modelling, profiling, debug, and code optimization or architectural knowledge of CPU and GPU.
Nice to have
- Python experience.
- MLIR experience.
- GPU programming experience (CUDA or OpenCL).
Skills: deep learning kernels, tile-based GPU programming, performance optimization, CUDA, MLIR
This role has been open 0 days — well below the 55-day median for Infrastructure/Platform Software roles.
Infrastructure/Platform Software · GPU Software Stack
|
Open roles in category
220
|
Median days open
55 d
|
Median salary
$236k
|
See the full market breakdown ▾Category comparison, skills in demand, and who else is hiring
| Metric | NVIDIA | All employers we track in this specialty (220 roles · 19 employers) |
|---|---|---|
| Open roles in this specialty | 123 | 220 |
| Open roles in the wider Software, Firmware & Systems family | 961 | 5961 · 145 employers |
| Median days open | 49 d | 55 d (−6 d vs this employer) |
| Median salary (USD postings) | — | $236k |
Skills observed across this category: deep learning kernels, tile-based GPU programming, performance optimization, CUDA, MLIR
Who's hiring in this category
- NVIDIA (this employer) · 120 open roles · median 51 d
- AMD · 51 open roles · median 51 d
- Qualcomm · 19 open roles · median 160 d
- Intel Corporation · 6 open roles · median 77 d
- Arm Holdings · 5 open roles · median 98 d
- d-Matrix · 2 open roles · median 106 d
How we counted: 220 open Infrastructure/Platform Software (GPU Software Stack) roles from 19 employers tracked in the SemiconductorJobs index, counted 1 Oct 2026. Specialty figures count only roles carrying this exact specialty label, so an employer's related work in neighbouring specialties is not included there — it is counted in the wider Software, Firmware & Systems family row. Figures refresh nightly.