NVIDIA Developer

# NVIDIA Developer 
Comprehensive developer portal for NVIDIA’s accelerated computing and AI tools. 


## Generative AI
Create scalable generative AI solutions using neural networks to learn patterns from existing data and generate new, original text, image, audio, and video content.
- [NeMo Customizer](https://developer.nvidia.com/nemo-customizer.md): Fine-tune LLMs using supervised techniques
- [NeMo Evaluator](https://developer.nvidia.com/nemo-evaluator.md): Comprehensive evaluation capabilities for LLMs
- [NeMo Guardrails](https://developer.nvidia.com/nemo-guardrails.md): Safety checks and content moderation
- [NeMo Agent Toolkit](https://developer.nvidia.com/nemo-agent-toolkit.md): Build AI-powered conversational agents and agentic applications with NeMo
- [NeMo Retriever](https://developer.nvidia.com/nemo-retriever.md): Multimodal retrieval-augmented generation microservices
- [NIM](https://developer.nvidia.com/nim.md): Inference microservices for foundation models


## Inference Optimization
Deploy high-performance AI inference workloads.
- [TensorRT](https://developer.nvidia.com/tensorrt.md): Ecosystem of APIs, compilers, and runtimes for high-performance deep learning inference
- [Dynamo](https://developer.nvidia.com/dynamo.md): Unified framework for high-performance LLM inference with KV-aware routing and SLA-based auto-scaling


## Data Science
Analyze large-scale data with GPU-accelerated libraries for machine learning and analytics.
- [CUDA-X Data Science](https://developer.nvidia.com/topics/ai/data-science/cuda-x-data-science-libraries.md): High-performance GPU-accelerated suite for modern data science workflows
- [cuDF](https://developer.nvidia.com/topics/ai/data-science/cuda-x-data-science-libraries/cudf.md): GPU DataFrame library accelerating pandas workflows
- [cuML](https://developer.nvidia.com/topics/ai/data-science/cuda-x-data-science-libraries/cuml.md): GPU-accelerated machine learning algorithms compatible with scikit-learn
- [NeMo Curator](https://developer.nvidia.com/nemo-curator.md): High-speed, scalable data curation and preparation for AI training
- [Morpheus](https://developer.nvidia.com/morpheus-cybersecurity.md): End-to-end AI pipeline for cybersecurity analytics and processing
- [cuVS](https://developer.nvidia.com/cuvs.md): GPU-accelerated vector search and clustering to supercharge search and RAG


## Healthcare
Accelerate drug discovery, medical imaging, and clinical AI development with NVIDIA healthcare platforms.
- [BioNeMo](https://www.nvidia.com/en-us/clara/biopharma/): Generative AI platform and SDK for chemistry, biology, and drug discovery
- [Clara Guardian](https://developer.nvidia.com/clara-guardian.md): Application framework for building and deploying smart sensors and multimodal AI in healthcare facilities
- [Isaac for Healthcare](https://developer.nvidia.com/isaac/healthcare.md): Robotics development platform for building AI-powered surgical, diagnostic, and medical automation systems
-[HoloScan](https://developer.nvidia.com/holoscan-sdk.md): Accelerate AI-powered medical and sensor device development with real-time edge data processing


## Quantum Computing
Simulate quantum circuits and develop hybrid quantum-classical solutions on NVIDIA GPUs.
- [cuQuantum](https://developer.nvidia.com/cuquantum-sdk.md): Libraries and tools for quantum circuit and device-level simulation
- [CUDA-Q](https://developer.nvidia.com/cuda-q.md): Open-source quantum development platform for hybrid systems
- [CUDA-QX](https://developer.nvidia.com/cuda-qx.md): Extension for advanced quantum simulations
- [cuPQC](https://developer.nvidia.com/cupqc.md): Toolkit for pre-quantum computing R&D


## CUDA
Develop GPU-accelerated applications.
- [CUDA Toolkit](https://developer.nvidia.com/cuda-toolkit.md): Complete development environment for building GPU-accelerated applications


## CUDA-X Libraries
Accelerate core computing and domain-specific workloads with NVIDIA-optimized CUDA-X software libraries.
- [cuBLAS](https://developer.nvidia.com/cublas.md): Basic Linear Algebra Subprograms
- [cuDNN](https://developer.nvidia.com/cudnn.md): Deep neural network library
- [cuFFT](https://developer.nvidia.com/cufft.md): Fast Fourier Transform library
- [cuPyNumeric](https://developer.nvidia.com/cupynumeric.md): NumPy replacement
- [NVPL](https://developer.nvidia.com/nvpl.md): NVIDIA Performance Libraries
- [Thrust](https://developer.nvidia.com/thrust.md): A parallel algorithms library for C++ that resembles the C++ Standard Template Library (STL).
- [CUB](https://docs.nvidia.com/cuda/cub/index.html): A reusable software components library for building high-performance CUDA kernels.
- [cuSOLVER](https://developer.nvidia.com/cusolver.md): A GPU-accelerated library for dense and sparse direct solvers.
- [cuSPARSE](https://developer.nvidia.com/cusparse.md): A GPU-accelerated library for basic linear algebra subroutines with sparse matrices.
- [cuTENSOR](https://developer.nvidia.com/cutensor.md): A GPU-accelerated tensor linear algebra library.
- [cuDSS](https://developer.nvidia.com/cudss.md): GPU-accelerated Direct Sparse Solver library for solving linear systems with very sparse matrices.
- [cuRAND](https://developer.nvidia.com/curand.md): Delivers high performance GPU-accelerated random number generation (RNG).
- [Nvmath-python](https://developer.nvidia.com/nvmath-python.md): An open source library that gives Python applications high-performance pythonic access to the core mathematical operations implemented in the CUDA-X Math Libraries.
- [cuEquivariance](https://developer.nvidia.com/cuequivariance.md): Provides optimized NVIDIA CUDA kernels and comprehensive APIs, including those for triangle attention and triangle multiplication, to accelerate geometry-aware neural networks.
- [cuLitho](https://developer.nvidia.com/culitho.md): A library for accelerating computational lithography and the manufacturing process of semiconductors.
- [Warp](https://developer.nvidia.com/warp-python.md): Open-source kernel-based spatial computing library for Python, enabling GPU-accelerated simulation, differentiable programming, and advanced data generation for ML, robotics, and Omniverse digital twins


## Multi-GPU and Multi-Node Communication
Build scalable, high-performance applications that coordinate data exchange and computation across multiple GPUs and systems with NVIDIA’s communication libraries.
- [NCCL](https://developer.nvidia.com/nccl.md): Implement fast, topology-aware collective and point-to-point communication for multi-GPU and multinode systems
- [NVSHMEM](https://developer.nvidia.com/nvshmem.md): Enable scalable, efficient one-sided and GPU-initiated communication with a partitioned global address space for multi-GPU clusters
- [GPUDirect Storage](https://developer.nvidia.com/gpudirect-storage.md): Direct GPU-storage path
- [Magnum IO](https://developer.nvidia.com/magnum-io.md): Unify networking, storage, and compute IO management for large multi-GPU, multi-node data centers at scale
- [Legate](https://docs.nvidia.com/legate/latest/index.html): Distributed programming framework


## Networking
Build advanced, high-performance data center networks and communication frameworks with NVIDIA’s accelerated networking platforms, SDKs, and communication libraries.
- [NVIDIA Networking Platforms](https://developer.nvidia.com/networking.md): Solutions for InfiniBand and Ethernet connectivity, smart DPUs, and networking hardware integration for AI, HPC, and data analytics at massive scale.
- [DOCA Software Framework](https://developer.nvidia.com/networking/doca.md): SDK and runtime for developing software-defined, secure, GPU- and DPU-accelerated networking, storage, and security services across data centers.
- [NVIDIA Aerial](https://developer.nvidia.com/aerial.md): Platform and tools to build accelerated, software-defined 5G/6G radio access networks and wireless AI systems leveraging GPUs and DPUs.
- [HPC-X](https://developer.nvidia.com/networking/hpc-x.md): Complete communication stack with MPI, SHMEM, PGAS libraries, and performance-boosting collectives for InfiniBand- and Ethernet-enabled HPC clusters.
- [Magnum IO](https://developer.nvidia.com/magnum-io.md): Developer SDK for optimizing I/O and communication for AI, HPC, data science, and visualization—supporting storage, network, and GPU data movement at scale.
- [Rivermax](https://developer.nvidia.com/networking/rivermax.md): Optimized IP-based SDK for high-throughput, low-latency media and data streaming, SMPTE 2110 compliance, and direct NIC-to-GPU transfer for video and sensor applications.
- [InfiniBand](https://developer.nvidia.com/networking/infiniband-software.md): High-speed interconnect
- [Ethernet Switch SDK](https://developer.nvidia.com/networking/ethernet-switch-sdk.md): Develop high-performance, programmable Ethernet switches with advanced routing, switching, and abstraction APIs


## Game Development and Graphics SDKs
Render photorealistic visuals and accelerate ray-traced graphics with NVIDIA’s industry-leading real-time ray tracing toolkits.
- [CloudXR SDK](https://developer.nvidia.com/cloudxr-sdk.md): XR streaming platform
- [RTX Kit](https://developer.nvidia.com/rtx-kit.md): Suite of neural rendering and ray tracing technologies for real-time, photorealistic graphics and advanced game development
- [DLSS](https://developer.nvidia.com/rtx/dlss.md): Deep Learning Super Sampling to boost frame rates and image quality using AI-powered upscaling
- [OptiX](https://developer.nvidia.com/rtx/ray-tracing/optix.md): GPU-accelerated ray tracing engine and programming framework for rendering and visualization
- [ACE](https://developer.nvidia.com/ace-for-games.md): Avatar Cloud Engine for building AI-powered, conversational digital characters in games
- [NVIDIA VRWorks™ Graphics](https://developer.nvidia.com/vrworks.md): Graphics APIs and tools for building industry-leading virtual reality experiences