Role involves integrating new CUDA features and runtime abstractions into AI frameworks, performing workload and performance analysis, collaborating with researchers and architects, and developing runtime/tools for multi-GPU and multi-node deep learning workloads. Requires 8+ years industry experience, strong skills in CUDA, C++, Python, deep learning frameworks, distributed ML techniques, and performance profiling/benchmarking.

Work on research-oriented projects that bridge deep learning frameworks and low-level CUDA to optimize performance from single GPUs to cluster-scale supercomputers. Responsibilities include designing and optimizing custom CUDA kernels, architecting distributed compute systems, profiling hardware-software interactions, and collaborating with researchers and compiler/driver teams. Requires 8+ years experience and strong background in deep learning (transformers), systems programming, and GPU performance optimization.

NVIDIA seeks a Principal Software Engineer for DGX Cloud to build and operate production GPU clusters for AI workloads. Responsibilities include Kubernetes-based GPU resource scheduling, operator development, node health monitoring, telemetry and diagnostics ingestion, incident management, and improving reliability and performance of large-scale systems. Requires strong software engineering experience (15+ years), proficiency in Go or Python, and deep knowledge of distributed systems and cluster management.

NVIDIA seeks a Senior AI Infrastructure Software Engineer for the DGX Cloud team to design, build, and maintain scalable AI platforms for large-scale training, inference, fine-tuning, and agentic AI. Responsibilities include developing platform tools, optimizing AI/ML workload efficiency and resiliency, triaging failures from application to hardware, co-designing APIs, and defining reliability metrics. Requires 8+ years building software infrastructure for large-scale AI systems, experience with Kubernetes and observability stacks (ELK, Prometheus, Loki), proficiency in Golang/Python/C/C++, and experience with distributed systems and AI training/inference. Location: US (remote / Redmond, Seattle, Santa Clara).

Work on the open-source Triton compiler to design and implement compiler technology (using MLIR and Triton's Python DSL), optimize high-level kernels to efficient low-level GPU code, and hand-tune critical paths (inline PTX). Collaborate with hardware architects and the CUDA compiler team; experience tuning BLAS/deep-learning kernels, numerics/linear algebra, and ML compilers (TVM/MLIR) is a plus. Base salary range: 184,000 USD - 287,500 USD per year.

Join NVIDIA's Compute Compiler Team to upstream features and infrastructure from downstream repositories into LLVM, Clang, MLIR and related projects. Requires 6+ years working with LLVM/Clang/MLIR or similar, strong C++ skills, deep compiler internals knowledge, and experience engaging in open-source contribution workflows.

Role focuses on building performance testing frameworks, automated CI/CD pipelines, benchmarking systems, and dashboards to track and analyze GPU and compiler performance across models and hardware. Requires 5+ years software engineering experience, strong Python (C++ a plus), familiarity with deep learning frameworks (PyTorch/TensorFlow/JAX/TensorRT), GPU/hardware-aware performance analysis, profiling, and debugging across software and hardware layers. Base salary range listed: 152,000 USD–241,500 USD.

NVIDIA seeks a Senior DL Algorithms Engineer to enable and optimize open LLM/Omni models (e.g., Nemotron, Cosmos) on NVIDIA's accelerated inference software stack. Responsibilities include contributing to open-source frameworks (TRT-LLM, vLLM, SGLang, FlashInfer), profiling and removing bottlenecks across the inference stack, benchmarking, and co-designing with partner teams. Requires 3+ years' experience, strong deep learning and inference background, GPU performance optimization, PyTorch or equivalent, and understanding of computer/GPU architecture.

Build and optimize inference stacks (vLLM), develop and tune GPU kernels, extend compiler/DSL infrastructure, define benchmarking (including MLPerf), and architect containerized large-scale inference deployments. Requires strong Python and C/C++ skills, GPU/CUDA experience, familiarity with PyTorch and inference engines, container/orchestration experience, and advanced performance engineering; typical experience expectations are 7+ years (BS) or 5+ years (MS).

NVIDIA is hiring for its Deep Learning Compiler (DLC) team to develop compiler optimization algorithms, performance optimizations, and compiler infrastructure for neural networks used across data centers, devices, automotive, and robotics. The role requires 3+ years of relevant work or research experience, proficiency in C/C++ and Python, and familiarity with technologies such as CUDA/OpenCL, MLIR, XLA, TVM, LLVM, and deep learning frameworks like PyTorch. Base salary range: 152,000 USD - 241,500 USD per year.

NVIDIA seeks a Senior Software Developer for AI Networking to develop production communication frameworks, tools and benchmarks, enable AI models in benchmarking infrastructure, and design automation systems for large-scale clusters. Requires B.Sc./M.Sc. in CS or Software Engineering and 5+ years' experience, professional Python development skills, and strong Linux/command-line expertise. Experience with PyTorch/LLMs, Slurm/Kubernetes, MPI/HPC, InfiniBand/Ethernet, and performance optimization is desirable.

NVIDIA is hiring a GPU Compiler Engineer to understand, modify, and improve a proprietary C++ GPU compiler backend and design new compiler passes and optimizations. The role requires 2+ years of compiler code generation experience, strong C++ skills, and collaboration with global compiler, hardware, and application teams; experience with LLVM, CUDA, and graphics APIs is listed as advantageous.

Role involves evolving PTX language and PTXAS compiler infrastructure, collaborating with GPU architecture and CUDA programming-model teams, implementing compiler front end, optimizer/runtime interactions, and support for object files, debug info, linkers/loaders. Requires BS/MS/PhD in CS or related, 6+ years compiler experience, strong C/C++ skills, and familiarity with GPU ISAs and compute APIs like CUDA.

NVIDIA is hiring a Principal Systems Software Engineer to work on custom CEF and Chromium code for GeForce NOW client platforms. Responsibilities include improving performance, reliability, cross-platform compatibility, security, CI/CD and patching infrastructure, and upstreaming applicable changes. The role requires 15+ years' experience, strong C++ and JavaScript skills, and a history of contributions to Chromium or CEF repositories.

Work on NVIDIA's proprietary GPU compiler backend (C++): design and implement register allocation and other code-generation passes, collaborate with hardware and application teams, and improve compiler performance and robustness. Requires 3+ years of compiler code-generation experience (LLVM preferred); base salary range listed as 152,000–241,500 USD per year.

Develop and optimize GPU-accelerated AI inference serving software, drive convergence of Triton and Dynamo stacks, and contribute to open source. Requires MS/PhD or equivalent experience, 5+ years on deep learning software, strong Rust and C++ skills, familiarity with Python, and experience with high-scale distributed and ML systems.

Work on architecting and building reinforcement-learning post-training infrastructure that scales from single-GPU experimentation to production across thousands of nodes. Responsibilities include optimizing training-inference-rollout loops on GPUs/CPUs/LPUs, improving RL frameworks (e.g., VeRL, Miles, TorchTitan), ensuring fault tolerance and elastic scaling, and collaborating with researchers and hardware/software teams. Requires MS/PhD or equivalent experience and 5+ years in distributed systems, HPC, deep learning infrastructure, or ML systems engineering.

NVIDIA Aerial team seeks a senior engineer to architect and implement a GPU ray-tracing engine that produces volumetric radio maps and millisecond‑cadence per-link multipath channel updates. Role requires a PhD (or equivalent), 8+ years relevant experience, hands‑on CUDA and GPU ray‑tracing framework experience (OptiX, Vulkan RT, Embree), proficiency with GPU-friendly spatial data structures (BVH, space-filling curves, hash maps), and working knowledge of electromagnetic propagation. Salary ranges provided (USD, annual).

Lead the full DGX platform delivery lifecycle (firmware, OS, drivers, CUDA, networking, validation and GA release), manage vendors, define validation strategy, drive platform architecture and bring-up, and build and mentor an engineering organization. Requires 12+ years in systems firmware/software engineering and 5+ years in engineering leadership.

Responsible for developing and maintaining continuous development systems and tooling across the software stack (from user-facing frameworks to compilers and drivers). Requires MS/PhD or equivalent experience, 5+ years working on large software projects, proficiency in C++ and/or Python, scripting, Linux, and experience with CI/CD and cloud-hosted developer environments.