Perplexity
Member of Technical Staff (AI Inference Engineer)
$220K–$485K
San Francisco +2 · Staff · Full-time · Posted Apr 13, 2026
- Location
- San Francisco
Palo Alto
New York City - Experience
- 3+ years of professional software engineering experience with meaningful work on ML inference or high-performance systems.
About the job
Build and operate Perplexity’s high-performance inference infrastructure for AI models, optimizing GPU kernels, serving systems, performance, and reliability.
What you'll do
- Support transformer-based retrieval, text-generation, and multimodal models in production inference infrastructure.
- Port CUDA kernels to CuTe DSL and develop the Rust-based inference server.
- Optimize performance and build reliability and observability systems; respond to production incidents.
What they ask for
- GPU programming and performance experience with CUDA, Triton, CUTLASS, or similar.
- Ability to bring modern LLM architectures up reliably in production.
- Experience building and operating production distributed systems under load.
- At least 3 years of professional software engineering experience with meaningful work on ML inference or high-performance systems; familiarity with a deep learning framework and GPU architectures.
Nice to have
- Experience with ML compilers/framework internals, distributed GPU communication, low-precision inference, profiling tools, or Kubernetes.
- Deep systems programming experience.
Pay
$220K – $485K USD base salary
Tools
Rust, Python, CUDA, CuTe DSL, Triton, CUTLASS, PyTorch, JAX
AI focus
Develop and optimize inference infrastructure for transformer-based retrieval, text-generation, and multimodal models, including model serving and GPU kernels.
Model-written summary · the original posting is authoritative