Senior System Software Engineer - Dynamo-Triton Inference Server

NVIDIA is hiring a Senior System Software Engineer for its GPU-accelerated deep learning software team. The role focuses on developing and advancing the Dynamo-Triton Inference Server, driving the convergence of the Triton and NVIDIA Dynamo stacks, and building a high-performance AI inference platform for LLM and non-LLM workloads.

Responsibilities:

• Develop world-class GPU-accelerated AI inference serving software
• Contribute to feature development and drive broad customer adoption
• Drive the convergence of the Triton Inference Server and NVIDIA Dynamo stacks to establish a unified, high-performance inference platform. This platform will ensure feature parity and effectively serve both Large Language Model (LLM) and non-LLM workloads
• Be an active member of the open source deep learning software engineering community
• Balance a variety of objectives such as building robust software designed to be deployed in production server or cloud environments, optimizing and balancing prediction throughput and latency, and developing and adopting the next generation of inference technologies

Requirements:

• MS or PhD in Computer Science or relevant field (or equivalent experience)
• 12+ years of professional experience working on deep learning software
• Excellent Rust & C++ skills, familiarity with Python, and strong programming & software design skills including debugging, performance analysis, and test design
• Experience with high-scale distributed systems and ML systems
• Strong communication skills and ability to work in a fast-paced, agile team environment
• Prior experience with AI frameworks and engines, such as TensorRT, PyTorch, ONNX, OpenVINO, vLLM, or TRT-LLM
• Knowledge of GPU memory management, cache management, or high-performance networking
• Experience with distributed systems programming
• Experience in contributing to a large open source project: use of GitHub, bug tracking, branching and merging code, OSS licensing issues handling patches, etc