thinQmaster Virtual testing partner Adaptive practice exams that find your weak spots and drill them. Every answer explained. From $69
Enroll

thinQmaster

Find your exam

Five questions free, before you decide.

Step 1 · choose a vendor

In one of our camps? Your practice exams come with the tuition — they are on your study page already.

Home › Courses › NVIDIA › NVIDIA-Certified Professional: AI Infrastructure (NCP-AII)
NVIDIANVNCPAII

NVIDIA-Certified Professional: AI Infrastructure (NCP-AII)

We do not teach just for the test.Worked examples and real discussion, not a question bank. The exam is then delivered in the classroom — 98% pass it first time.
$2,795catalog price — your starting point
thinQflex™Tell us what you would like to pay for this class — in full, or across twelve monthly payments.
Next date
  • 4 Jan 2027Sterling, VA
Make your offerCourse datasheet (PDF)
4 daysconsecutive days
Live onlineor in the classroom
ProfessionalArtificial Intelligence
50% labs50% lecture
24 monthsfree retakes

Course overview

Download the course datasheetThe full outline for NVIDIA-Certified Professional: AI Infrastructure (NCP-AII), as a PDF.
thinQtank® Learning is offering a unique four-day training camp comprised of four days of instructor-led learning that gives students the knowledge and skills needed to bring up GPU servers and systems, manage the physical layer, install and configure the control plane, test and verify AI clusters, and troubleshoot and optimize AI infrastructure.

The camp follows the official NVIDIA NCP-AII exam topics and study guide, with hands-on labs bringing up, validating and tuning GPU systems and clusters throughout.

NVIDIA, CUDA, RAPIDS, InfiniBand, Spectrum, NeMo, NIM and Omniverse are trademarks of NVIDIA Corporation. OpenUSD is maintained by the Alliance for OpenUSD.

This course prepares students to take the NVIDIA-Certified Professional: AI Infrastructure (NCP-AII) exam.

What's included

13 items
thinQtank® Learning Accelerated Custom Exam Study Guide for NCP-AII
Hands-on labs bringing up, validating and tuning GPU systems and clusters
Custom and focused exam preparation software and materials
Course specific thinQtank® Learning publications to promote a fun and exciting learning
Membership to the thinQtank® Learning University Online
WE DO NOT "TEACH THE TEST" - Students will receive valuable examples and discussion
All courses taught by AI engineers experienced with NVIDIA platforms
Additional hours of remote lab access
Custom video recordings with even more in-depth learning of course topics
Customized practice exam software
Digital courseware
Retake any or all portions of the course as many times as you like in person or live on-line for 24 months
Six months mentoring access to our engineers after completing the course

What you'll be able to do

20 objectives
  • Bring up GPU servers: BMC, OOB and TPM setup, firmware upgrades (including HGX), power and cooling validation
  • Install and validate GPUs, cables, transceivers and third-party storage
  • Configure BlueField network platforms and MIG for AI and HPC
  • Install and configure Base Command Manager with HA, Slurm, Enroot and Pyxis
  • Install GPU and DOCA drivers, the NVIDIA Container Toolkit and NGC CLI
  • Verify clusters with stress tests, HPL, NCCL, ClusterKit and burn-in
  • Diagnose and replace faulty GPUs, NICs, fans and PSUs and tune server and storage performance
  • AI Factory Architecture and Network Topologies
  • Deployment and Validation Sequence
  • BMC, Out-of-Band and TPM Initial Setup
  • Firmware Upgrades and Fault Detection (incl. HGX)
  • GPU Server Installation, Cabling and Transceivers
  • Physical Layer: BlueField and MIG Configuration
  • Base Command Manager Installation and HA
  • Cluster Setup: Slurm, Enroot and Pyxis
  • GPU and DOCA Drivers, Container Toolkit and NGC CLI
  • Single-Node Testing: Stress, HPL and NCCL
  • Fabric Verification: Cable Signal Quality, Firmware and NCCL East-West
  • Burn-In and Storage Testing
  • Troubleshooting and Optimization

The exam

75questions maximum
120minutes to complete
  • Five topic areas: System and Server Bring-up (31%); Physical Layer Management (5%); Control Plane Installation and Configuration (19%); Cluster Test and Verification (33%); Troubleshoot and Optimize (12%)
  • Delivered online with a remote proctor through Certiverse; English only
  • Exam fee: $400 (set by NVIDIA)
  • Valid two years; recertify by retaking the current exam

Who it's for, and what you need first

Target Audience

  • Data Center Administrator
  • Infrastructure / System Administrator
  • Network Engineer
  • Solution Architect
  • Service members, veterans and military spouses using AF COOL, Army Credentialing Assistance, MyCAA or VET TEC 2.0

Prerequisites

  • No certification is required
  • Recommended: Two to three years of data center operations with NVIDIA hardware
  • NCA-AIIO or equivalent data center experience is a strong foundation

How it's delivered

Instructor-Led Training
Immersive Live-Online Training
On-Site and Custom Delivery

Dates and locations

20 scheduled
Guaranteed to run means the class goes ahead whatever the enrollment.
Prices shown are MSRP. Qualified military personnel funding through Air Force COOL, Army Credentialing Assistance, MyCAA or VET TEC 2.0 pay a flat $1,500 per class, whatever the listed price. Everyone else — call 855-868-4467 about group pricing, discounts and current promotions.