NVIDIA Corporation
Official@nvidia · 2788 San Tomas Expressway, Santa Clara, CA, 95051
Offers high-performance computing, distributed training, and GPU-accelerated infrastructure management for enterprise-scale machine learning and simulation workloads.
Agent Skills by NVIDIA Corporation
Showing 630 vetted skills indexed across 25 GitHub repositories.
NVIDIAtilegym-converting-cutile-triton-to-cutile-rs
Convert Triton-TileIR and cuTile-Python GPU kernels to cutile-rs Rust via a multi-agent pipeline.
NVIDIAcuopt-skill-evolution
Detect generalizable learnings from problem-solving and propose skill updates.
NVIDIAdeveloper-bump-version
Bump the Earth2Studio package version and insert a blank CHANGELOG section on main.
NVIDIAnvidia-skill-finder
Discover and recommend NVIDIA agent skills from the live catalog for NVIDIA-related tasks.
NVIDIAdoca-comm-channel-admin
Enumerate host-DPU DOCA comch servers and connections via the doca_comm_channel_admin binary.
NVIDIAevent-video-generation-workflow
Run the PAIDF Event Video Generation Airflow DAG on Kubernetes for anomaly video synthesis and auto-labeling.
NVIDIAdynamo-interconnect-check
Validate NIXL/UCX/NCCL interconnect readiness for disaggregated Dynamo serving over RDMA and NVLink.
NVIDIAnemo-mbridge-perf-moe-long-context
Guides context parallelism sizing and recompute strategies for long-context MoE training with Megatron Bridge.
NVIDIAdicom-series-to-volume
Converts a single CT DICOM series folder into a HU-scaled NIfTI volume with affine geometry.
NVIDIAnemotron-speech
Deploys, runs, and tests NVIDIA Riva ASR, TTS, and NMT NIMs on cloud or self-hosted infrastructure.
NVIDIAnemo-mbridge-perf-parallelism-strategies
Guides selection and sizing of TP, PP, DP, CP, and EP parallelism configurations in Megatron Bridge.
NVIDIAearth2studio-create-prognostic
Create Earth2Studio prognostic model wrappers for time-stepping weather forecasts.
NVIDIAi4h-workflow-dataset-mimic
Expand HDF5 robot demonstration recordings by cloning trajectories with action and state noise.
NVIDIAskill-card-generator
Generate governance skill cards for existing agent skill directories using Jinja templates.
NVIDIAtao-train-grounding-dino
Train, evaluate, export, and deploy TAO Grounding DINO models for text-prompted open-set object detection.
NVIDIAtao-port-huggingface-model
Integrate HuggingFace computer vision models into the NVIDIA TAO Toolkit training and TensorRT deployment pipeline.
NVIDIAnv-generate-mr
Generate synthetic body MRI volumes using the NV-Generate-CTMR rflow-mr diffusion pipeline.
NVIDIAorchestration-setup
Audit, prepare, and deploy PAIDF Orchestration on Kubernetes GPU clusters.
NVIDIAjetson-customize-fan
Edits nvfancontrol fan profiles and boot defaults in Jetson BSP overlay configurations.
NVIDIAjetson-promote-image
Promote overlay files and built artifacts into a staged Jetson BSP image.
NVIDIAnemo-relay-install
Installs NeMo Relay CLI, language packages, and framework integrations with verified setup checks.
NVIDIAjetson-link-docs
Registers pre-downloaded Jetson reference document paths into the active target-platform profile YAML.
NVIDIAtao-run-on-kubernetes
Submits TAO container jobs as Kubernetes Jobs with NVIDIA GPU scheduling.
NVIDIAdoca-setup
Verifies, configures, and troubleshoots the NVIDIA DOCA install and runtime environment on BlueField hosts.
Frequently Asked Questions About NVIDIA Corporation
FAQPage SchemaWhat specific tasks can engineers perform using these capabilities?▼
Engineers can execute distributed training on SLURM clusters, profile GPU kernel bottlenecks using Nsight, manage Kubernetes networking for high-performance clusters, and deploy quantized models via vLLM or SGLang endpoints.
Which technical personas are these capabilities designed for?▼
These capabilities target infrastructure engineers, machine learning researchers, performance optimization specialists, and network architects focused on high-throughput GPU computing and large-scale model deployment.
What are the prerequisites for running these distributed training workloads?▼
Workloads require NVIDIA GPU hardware, CUDA-compatible drivers, and specific environment configurations including NCCL for collective communication, SLURM for job scheduling, and containerized dependencies managed via uv or Docker.