Skip to content
MenuClose
All 3 NVIDIA exams

NVIDIA exam

NCA-GENL Exam facts

NVIDIA-Certified Associate: Generative AI LLMs

Questions
50-60
Minutes
60
Exam fee
$125
Scoring (no score shown)
Pass/fail

Independent: not affiliated with NVIDIA. These exam facts are free; the study bundle adds a study guide and two practice exams.

Short answer

NCA-GENL is the exam for the NVIDIA-Certified Associate: Generative AI LLMs certification, for associate developers who build applications on large language models with NVIDIA software. The exam has 50-60 questions in 60 minutes and costs $125.

Sources: exam page, certification FAQ

Published NVIDIA facts for NCA-GENL on one page. Last updated

Study bundle

Prepare for NCA-GENL

$19
Buy now - $19

Secure checkout, in a new tab.

Your receipt has a license key. Enter it at overqual.com/download for the three PDFs, stamped with your email and order number. They are yours to keep.

Try 5 free sample questions first See 3 pages inside
  • A 60-page study guide: every exam domain, weighted like the exam
  • Two practice exams: 120 original questions, every answer explained
  • Every bundle includes 6 months of access: download the PDFs, and any new editions, until 6 months after your purchase date.

Edition 1.0, October 2026

How many questions and how long?

The NCA-GENL exam has 50-60 multiple-choice questions in 60 minutes. NVIDIA does not say whether any of them are unscored.

Source: NVIDIA exam page

The exam

Questions
50-60
Time
60 minutes
Language
English

The sitting

CERTIVERSE
Booking
Certiverse account
Breaks
None when remotely proctored
Per sitting
One exam

Question types

Multiple choice

The exam page describes the exam as multiple-choice questions.

Pass or fail

NVIDIA's certification FAQ says you will not receive a score.

Unscored questions

NVIDIA does not say whether any questions are unscored.

Sources: NVIDIA exam page, certification FAQ

What does it cost?

The NCA-GENL exam costs $125 per attempt. A retake after a failed attempt is a new purchase.

Exam fee $125

After a failed attempt you wait 14 days before the next one, with at most 5 attempts in 12 months.

Sources: NVIDIA exam page, certification FAQ

What score do you need to pass?

NVIDIA does not publish a passing score for NCA-GENL. The exam is pass or fail, and you will not receive a score.

Results usually appear on screen at the end and by email within 24 hours. The digital badge arrives by email, to accept on Credly.

Source: NVIDIA certification FAQ

What does the exam cover?

The NCA-GENL exam covers five domains. Core machine learning and AI knowledge is the largest, at 30% of the exam. Software development is 24%, experimentation 22%, data analysis 14% and trustworthy AI 10%.

Source: NVIDIA exam page. Domains are numbered in the order of the Overqual study guide, and task numbers follow the order of NVIDIA's bullets.

  • D130%

    Core Machine Learning and AI Knowledge

  • D222%

    Experimentation

  • D324%

    Software Development

  • D414%

    Data Analysis

  • D510%

    Trustworthy AI

30% / largest

Core Machine Learning and AI Knowledge

  1. 1.1Support deployment and evaluation: scale, performance, reliability
  2. 1.2Extract insight from large datasets
  3. 1.3Build RAG systems, chatbots and summarizers
  4. 1.4Curate and embed content for RAG
  5. 1.5Machine learning fundamentals
  6. 1.6Python NLP and data packages, vector databases
  7. 1.7Follow research papers on LLM trends
  8. 1.8Choose and use text embedding models
  9. 1.9Apply prompt engineering
  10. 1.10Traditional ML with spaCy, NumPy and Keras
Core ML on NVIDIA's exam page
22% of the exam

Experimentation

  1. 2.1Model training and training optimization
  2. 2.2Prepare data for pretraining, fine-tuning and RLHF
  3. 2.3Hardware and software tests for LLM applications
  4. 2.4Evaluate technologies: cost, portability, compatibility, usability
  5. 2.5Collect data from human subjects for RLHF
  6. 2.6Evaluate and refine models and benchmarks
  7. 2.7Run experiments on models and pipelines
Experimentation on NVIDIA's exam page
24% of the exam

Software Development

  1. 3.1Support deployment and evaluation: scale, performance, reliability
  2. 3.2Build RAG systems, chatbots and summarizers
  3. 3.3Python NLP packages and vector databases
  4. 3.4Identify the data, hardware and software components needed
  5. 3.5Monitor data collection, experiments and processes
  6. 3.6Traditional ML with spaCy, NumPy and Keras
  7. 3.7Write software components and scripts
Software Development on NVIDIA's exam page
14% of the exam

Data Analysis

  1. 4.1Extract insight from large datasets
  2. 4.2Compare models with statistical metrics
  3. 4.3Conduct data analysis under supervision
  4. 4.4Create graphs and charts
  5. 4.5Relationships, trends and confounding factors
Data Analysis on NVIDIA's exam page
10% of the exam

Trustworthy AI

  1. 5.1Ethical principles of trustworthy AI
  2. 5.2Data privacy and consent
  3. 5.3Technologies that improve trustworthiness
  4. 5.4Minimize bias in AI systems
Trustworthy AI on NVIDIA's exam page

Inside the study bundle

Study guide chapters

Eleven chapters. Chapters 3 to 8 follow the five exam domains and are weighted like the exam.

  1. 01 What the exam is p. 3
  2. 02 The NVIDIA LLM stack: a map of the names p. 8Foundation
  3. 03 How LLMs work: transformers, tokens, embeddings, decoding p. 14Domain 1 / 30%
  4. 04 Adapting LLMs: training, fine-tuning, prompting and RAG p. 22Domain 1 / 30%
  5. 05 Experimentation: training runs, data preparation, evaluation p. 29Domain 2 / 22%
  6. 06 Software development: building and deploying LLM applications p. 36Domain 3 / 24%
  7. 07 Data analysis: metrics, exploration and GPU data science p. 43Domain 4 / 14%
  8. 08 Trustworthy AI p. 47Domain 5 / 10%
  9. 09 Final review p. 51
  10. 10 Sources p. 58
  11. 11 Edition and updates p. 60

Look inside the study guide

Three pages of the Edition 1.0 study guide: the contents, its chart for choosing how to adapt a model and the first page of a domain chapter.

Page 2 of the study guide: the contents, eleven chapters with page numbers and domain weights, the four kinds of margin note and the companion practice exams
Study guide, page 2: Contents
Page 25 of the study guide: a decision chart that picks RAG, a better prompt, DPO or RLHF, LoRA or full fine-tuning from the data and compute you have, followed by a Mental model note
Study guide, page 25: Choosing how to adapt a model
Page 14 of the study guide: the first page of Chapter 3, How LLMs work: transformers, tokens, embeddings, decoding, with the chapter's introduction
Study guide, page 14: Chapter 3, How LLMs work: transformers, tokens, embeddings, decoding

What is free here, and what is in the bundle?

The exam facts, the change log and 5 sample questions are free on this site. The domain chapters, the NVIDIA stack map, the final review and the two 60-question practice exams are in the bundle.

Item Free on this site In the bundle
Exam facts and change log Yes, with links to NVIDIA sources Chapter 1 of the study guide sums up the exam's format and scoring.
5 sample questions Yes No. None of the 5 appear in the practice exams.
Domain chapters (chapters 3 to 8) No Yes, in the study guide
The NVIDIA LLM stack map and renamed products No Yes, chapter 2 of the study guide
Final review No Yes, chapter 9 of the study guide
Two 60-question practice exams with answer keys No Yes, Practice Exams 1 and 2

Questions before you buy

Is this official?

No, the NCA-GENL bundle is not official. Overqual is independent and is not affiliated with, sponsored by, or endorsed by NVIDIA Corporation. NVIDIA, NeMo, NIM, Triton, TensorRT, CUDA, RAPIDS, Dynamo and related names are trademarks of NVIDIA Corporation, used only to refer to the products and exam described. For official facts, read the NVIDIA exam page.

Does it include real exam questions?

No, the NCA-GENL bundle contains no exam questions, and every practice question is original.

How is the NCA-GENL bundle different from NVIDIA's free materials?

NVIDIA's free pages set out what the NCA-GENL exam covers. The NCA-GENL bundle adds a 60-page study guide that covers every exam domain, and two practice exams with 120 original questions, every answer explained. When NVIDIA revises the exam, a new edition follows if the changes warrant it, included while your access lasts.

How do I get updates?

Updates to the NCA-GENL bundle arrive as new editions. Every bundle includes 6 months of access: download the PDFs, and any new editions, until 6 months after your purchase date. Download each new edition at overqual.com/download with the license key from your receipt email. The change log on the exam facts page says what changed.

Can I get a refund?

No. Sales of the NCA-GENL bundle are final, because downloaded PDFs cannot be returned. Read the free exam facts and try the 5 free sample questions first. If a download fails or you paid twice by mistake, contact Overqual to have the download fixed or the second payment refunded.

Which exam version does it cover?

Edition 1.0 follows the NCA-GENL exam page as NVIDIA updated it on Sep 29, 2026, and NVIDIA documentation as of October 2026.

Ready for NCA-GENL?

Three PDFs, 105 pages: the study guide and two practice exams.

Buy now - $19 5 free sample questions

Which NVIDIA software does the exam name?

NVIDIA publishes no list of products in scope for NCA-GENL. The exam page's suggested reading names RAPIDS cuDF and cuML, TensorRT, Triton Inference Server and NeMo, next to open-source tools such as spaCy and Hugging Face.

Source: NVIDIA exam page, suggested reading

NVIDIA software in the reading

  • RAPIDS cuDFGPU DataFrames with a pandas-like API
  • RAPIDS cuMLGPU machine learning with a scikit-learn-like API
  • TensorRTCompiles a trained model into an optimized GPU engine
  • Triton Inference ServerServes many models and frameworks; now Dynamo-Triton
  • NeMoNVIDIA's libraries for training, customizing and guarding models

As of March 18, 2025, Triton Inference Server is part of NVIDIA Dynamo under the name Dynamo-Triton.

Named in the outline

  • spaCy
  • NumPy
  • Keras
  • Vector databases
  • Python NLP packages

Other tools in the reading

  • Hugging Face LLM Course
  • scikit-learn
  • Milvus
  • Matplotlib
  • Weights & Biases
  • LM Evaluation Harness
Out of scope NVIDIA publishes no in-scope or out-of-scope list for this exam. The domain bullets and the suggested reading on the exam page are the official outline.
RENAMED

Is Triton the same as Dynamo-Triton?

Yes, the Triton in the NCA-GENL suggested reading is now called Dynamo-Triton. The exam page's suggested reading still says Triton.

Exam page reading list

Links the Triton Inference Server quick start.

NVIDIA developer blog

Says that as of March 18, 2025, Triton is part of NVIDIA Dynamo, as Dynamo-Triton.

This site writes Dynamo-Triton (formerly Triton Inference Server).

How is it delivered?

You book the NCA-GENL exam through a Certiverse account, which the exam page links. NVIDIA's certification FAQ describes online exams with remote proctoring and in-person exams at authorized partners; the exam page does not say which apply, so check when you book. The exam is in English.

Booking
Certiverse account
Online
Remote proctoring, no breaks
Language
English
Prerequisites
A basic understanding of generative AI and LLMs
ID
Government-issued photo ID matching your registration

Sources: NVIDIA exam page, certification FAQ

Dates, retakes and validity

NVIDIA updated the NCA-GENL exam page on Sep 29, 2026 and published no change log. The five domains and their weights stayed the same. No retirement date is published.

Sep 29, 2026
Exam page updated: outline on the page, seven Experimentation bullets, two courses
Jan 2025
NVIDIA's earlier study guide PDF, no longer linked from the exam page
Booking
Up to 60 days ahead; cancel or reschedule at least 24 hours before
Retake
14 days after a failed attempt
Valid for
Two years from issuance; recertify by passing the exam again
Retirement
Not published

Sources: NVIDIA exam page, NVIDIA's study guide (PDF, Jan 2025), certification FAQ

Fact sheet

Show the 10 factsHide the facts
Who it is for
Associate developers who help build, program and test generative AI applications, under a senior team member. NVIDIA suggests a basic understanding of generative AI and LLMs. Source: Exam page
Format
50-60 multiple-choice questions in 60 minutes, in English Source: Exam page
Fee
$125 per attempt Source: Exam page
Scoring
Pass or fail. No passing score is published, and no score is reported. Source: Certification FAQ
Domain weights
Core ML and AI knowledge 30%, software development 24%, experimentation 22%, data analysis 14%, trustworthy AI 10% Source: Exam page
Delivery
Booked through a Certiverse account. Remotely proctored exams allow no breaks. Source: Exam page, certification FAQ
Retakes
14 days wait after a failed attempt; at most 5 attempts in 12 months. Source: Certification FAQ
Validity
Two years from issuance. Recertify by passing the exam again. Source: Exam page, certification FAQ
Recommended training
Fundamentals of Deep Learning (open-source course materials) and Rapid Application Development With LLMs Source: Exam page
Last page update
Sep 29, 2026. No change log published. Source: Exam page

What has changed, and when?

NVIDIA publishes no change log. Three dated changes so far, newest first: the exam page update and two renames of NVIDIA software.

  1. NVIDIA moved the outline onto the exam page, adding seven Experimentation bullets and cutting recommended training to two courses. NVIDIA exam page
  2. TensorRT Model Optimizer was renamed NVIDIA Model Optimizer. Model Optimizer on GitHub
  3. Triton Inference Server became Dynamo-Triton, part of the NVIDIA Dynamo platform; the exam page's reading list still says Triton. NVIDIA developer blog

To watch NVIDIA directly, check the NCA-GENL exam page.

NCA-GENL study bundle Buy now - $19