Join NVIDIA to enable CUDA Math Libraries (e.g., cuBLAS, cuSPARSE, cuSOLVER, cuFFT) on emerging and specialized platforms. Responsibilities include platform integration, defect triage, performance target identification and testing, and collaborating with CI/CD and build/test teams. Requires advanced degree or equivalent experience and 5+ years developing high-performance numerical software using C++ and Python, with strong numerical methods and parallel programming experience.

NVIDIA seeks a Senior System Software Engineer to develop and optimize Rust- and Python-based GPU-accelerated pipelines for indexing, deduplicating, filtering, and classifying multimodal datasets to support LLM/VLM training and Agentic Retrieval. The role requires 6+ years experience, hands-on MLOps and cloud familiarity (containers, Docker-Compose, Kubernetes), and expertise in LLM/VLM/Generative AI workflows, benchmarking, profiling, and performance optimization.

Lead technical vision and roadmap for 3D obstacle perception using multi-camera and multi-sensor fusion (camera, radar, lidar), design and productionize transformer- and CNN-based models (including BEV approaches), drive data and labeling strategy, define KPI frameworks, optimize for safety/latency/resource constraints, and mentor engineers. Requires 15+ years experience and proficiency with frameworks such as PyTorch, plus strong software skills in Python/C++ and CUDA.

Responsible for architecture, design, and implementation of production-grade 3D obstacle perception pipelines (multi-camera and multi-sensor fusion) using modern CNN and transformer techniques. Work includes model development, large-scale training and evaluation, data and labeling strategy, collaboration with safety/systems teams, and optimization for deployment (including CUDA/GPU work). Minimum experience ranges provided by degree (PhD 4+, MS 6+, BS 8+).

Role entails architecture and roadmap for radar-based 3D obstacle perception, research and model design (transformers, radar point clouds, BEV), sensor evaluation and integration, production-grade deep learning (PyTorch/CUDA), KPI/error analysis, and data/auto-labeling strategy. Requires 12+ years of relevant experience.

NVIDIA Learning and Perception research team seeks PhD interns to research and implement embodied and agentic AI methods (perception, agentic reasoning, multi-modal and continual learning, etc.). Interns collaborate with mentors and engineers, prototype on state-of-the-art hardware, and help transfer research to products. Must be pursuing a PhD in CS/EE or related field, have strong knowledge of computer vision/LLMs/agents, excellent programming skills (Python, C++), familiarity with ML frameworks such as PyTorch, and a promising research track record. Hourly pay range stated (USD 30–94 depending on location).

The role involves researching and developing techniques to GPU-accelerate deep learning, machine learning, and other AI/HPC workloads; performing in-depth analysis and optimization of complex algorithms on modern CPU/GPU architectures; collaborating with industry and academia; publishing/presenting optimization techniques; and influencing next-generation hardware and software design. Requirements include an advanced computational degree (or equivalent), 8+ years relevant experience, programming fluency in C/C++, experience with parallel programming models (CUDA, OpenACC, OpenMP, MPI, pthreads), low-level performance optimization skills, and strong CPU/GPU architecture knowledge.

Join NVIDIA DRIVE to design and implement synthetic-data and sensor-simulation tools (lidar, radar, camera/RGB-D) against NuRec and Cosmos environments, build world models, evaluate dataset quality and sim-to-real transfer, and operate large-scale simulation pipelines. Requires 8+ years of relevant experience, strong Python/C++ skills, and familiarity with ML workflows, profiling, and scalable engineering tools.

NVIDIA seeks a Senior Perception (Machine Learning) Engineer to research, develop, and productize end-to-end deep learning perception modules for traffic signals (lights, signs, road markings, construction objects) to enable autonomous driving without reliance on HD maps. Role requires strong hands-on deep learning and software skills (PyTorch, Python/C++), experience with data-driven development, and meeting product requirements for safety, latency, and robustness. Posted salary ranges: 184,000–287,500 USD (Level 4) and 224,000–356,500 USD (Level 5).

Responsible for full SDLC: design, implement, test, integrate and document perception software components and deploy state-of-the-art DNN models to embedded platforms. Requires 12+ years' experience with embedded real-time applications, strong C++ and Python skills, experience with test-driven development, and familiarity with CUDA/SIMD and deep learning frameworks (e.g., TensorFlow, PyTorch).

Join the DRIVE Road Structure / Online Mapping / Context Fusion team to design and implement learning-based, multimodal sensor-fusion systems and shared spatiotemporal world representations for autonomous vehicles. Requires 8+ years experience (including 2+ years in AV/robotics and 2+ years leadership), strong C++/Python skills, deep-learning experience (PyTorch), and expertise with LiDAR/camera/radar fusion, 3D perception, and GPU-accelerated training/inference.

Lead the architecture and implementation of LLM‑friendly math library APIs and documentation, collaborate with stakeholders to deliver LLM‑enhanced releases, build metrics and tooling to measure code‑generation impact, and prototype solutions to represent math library APIs for codegen and agent tooling. Preferred background includes MSc/PhD (or equivalent), 3+ years' experience, and familiarity with LLMs, API design, and CUDA‑X libraries.

NVIDIA GEAR is hiring a PhD research intern to design and implement AI algorithms and models for humanoid robots and embodied agents, develop large-scale training and inference methods, optimize and deploy models in simulation and on robot hardware, and collaborate with research and engineering teams. Required: pursuing a PhD in a related field, strong Python and prototyping skills, experience with ML frameworks (PyTorch/JAX/TensorFlow), and background in multimodal foundation models or robot learning. Expected hourly rate listed as 38–94 USD. Applications accepted through at least November 3, 2026.

Role at NVIDIA to drive architecture and implementation of math libraries APIs and documentation to be LLM-first, collaborate with stakeholders, build metrics and tooling for LLM/agentic code generation, and prototype solutions to make CUDA-X libraries accessible to LLMs and AI agents. Requires 3+ years' experience and strong knowledge of LLMs, API design, and CUDA/X math libraries.

Role focuses on turning research into production software for atomistic simulation workflows, collaborating with research, engineering, product teams and customers. Responsibilities include integrating material science applications with AI surrogates using NVIDIA ALCHEMI, profiling and benchmarking, collaborating on whitepapers and participating in workshops. Requires strong experience with HPC, Fortran, C++, CUDA, Python and materials science tools.

Role involves designing and implementing compiler transformations, developing MLIR-based dialects and lowering passes, and optimizing tile-based kernel performance across NVIDIA GPU architectures. Requires 3+ years in compiler optimization, IR design or performance analysis and strong C/C++ skills.

NVIDIA seeks a Principal EDA R&D Engineer to architect and implement advanced formal verification algorithms (model checking, SAT/SMT, BDDs, theorem proving) and develop multi-agent Agentic AI systems that scale formal tools to multi-billion-gate designs. Requires a Master's/Ph.D. or equivalent, minimum 15 years building commercial or advanced academic EDA/hardware verification tools, and expert C++ (C++17/20) and formal methods experience.

Lead the technical direction and implementation of synthetic data generation across text, code, structured, and multimodal domains to support pre-training, fine-tuning, and RL stages for frontier LLMs (e.g., Nemotron). Build scalable data pipelines, open-source libraries/SDKs, and privacy-preserving synthesis methods while collaborating with research, engineering, product, and external labs and publishing original research.

NVIDIA seeks a Senior MLOps Engineer to develop and deploy AI infrastructure solutions, advise customers and partners, profile and tune large-scale training and inference workloads, and build open-source tools and reference architectures. Requires 8+ years of technical experience, strong systems and programming skills (Linux, Kubernetes, Python, C++/Go/Rust), and experience with ML frameworks and datacenter-scale networking. Role is listed for Germany, Czechia, and Poland (remote options).

NVIDIA is hiring a Senior Software Engineer to build and operate scalable evaluation systems, sandboxed execution, model-provider integrations, and benchmarks for LLM agents. The role requires 8+ years of software engineering experience, expert Python skills, and working proficiency in Rust, Go, C++, or TypeScript, with responsibilities including diagnosing agent failures, designing controlled experiments, shipping harness/runtime improvements, and contributing upstream to open-source projects. The position is listed for US locations and remote.