AI Engineer
Core
Build and optimize inference systems for Zoom's AI models in production, focusing on latency, throughput, and cost reduction for transformer and multimodal models.
Role type
Senior IC AI Infrastructure Engineer (Inference Systems)
Builds
High-performance inference serving stack for Zoom's products
Domain
AI Infrastructure / High-Performance Computing / GPU Systems
Deliverable
production ML models
Required skills
Inference frameworks (vLLM, SGLang, TensorRT-LLM), GPU profiling and debugging, Python, C++, speculative decoding, continuous batching, KV cache management, quantization, distributed/multi-GPU serving
Preferred skills
CUDA/Triton kernel development, MoE or 100B+ parameter model serving, NCCL/RDMA/RoCE experience, ML systems research
Technologies
vLLM, SGLang, TensorRT-LLM, ONNX Runtime, Nsight Systems, PyTorch, CUDA, Triton, CUTLASS
Responsibilities
Build and tune inference serving components for transformer and MoE models; Implement inference optimizations (speculative decoding, batching, quantization); Profile GPU workloads and root-cause performance regressions; Extend inference frameworks for production requirements; Build internal benchmarking and performance modeling tooling; Partner with researchers to deploy new model architectures; Provide on-call support for production services
Seniority
Junior to Mid-level, hands-on IC
