NCP-AAI vs NCP-GENL vs NCA-GENL: Which NVIDIA AI Certification?

The short answer

NCP-GENL certifies the person who makes the model work — optimising it, fine-tuning it, getting it fast on GPUs and deploying it. NCP-AAI certifies the person who builds systems around models — agents that plan, use tools, remember, get evaluated and are kept safe in production. NCA-GENL is the associate exam underneath both: conceptual, an hour long, and the right starting point if neither professional exam describes your current work.

                 NCP-AAI              NCP-GENL               NCA-GENL
Full name ...... Professional:        Professional:          Associate:
                 Agentic AI           Generative AI LLMs     Generative AI LLMs
Level .......... Professional         Professional           Associate
Duration ....... 120 minutes          120 minutes            1 hour
Questions ...... 60-70                60-70                  50-60
Price (US) ..... $200                 $200                   $125
Valid for ...... 2 years              2 years                2 years
Domains ........ 10                   10                     5
Passing score .. not published        not published          not published

Where the professional blueprints actually differ

Put the two ten-domain blueprints next to each other and the overlap is smaller than the shared word "professional" suggests. NCP-GENL puts its weight on the model itself: Model Optimization 17%, GPU Acceleration and Optimization 14%, Prompt Engineering 13%, Fine-Tuning 13%, then Data Preparation and Model Deployment at 9% each, Evaluation and Production Monitoring and Reliability at 7% each, LLM Architecture 6% and Safety, Ethics, and Compliance 5%. NCP-AAI puts its weight on the system around the model: Agent Architecture and Design and Agent Development at 15% each, Evaluation and Tuning and Deployment and Scaling at 13% each, Cognition, Planning, and Memory and Knowledge Integration at 10% each, NVIDIA Platform Implementation 7%, and three 5% domains for running, safety and human oversight. Roughly 44% of NCP-GENL — optimisation, GPU acceleration and fine-tuning — has no direct counterpart in NCP-AAI, and the agent architecture, development, memory and planning domains that make up 40% of NCP-AAI have none in NCP-GENL.

Who each one expects

NVIDIA recommends two to three years of practical AI or ML experience with LLMs for NCP-GENL, naming transformer architectures, prompt engineering, distributed parallelism and parameter-efficient fine-tuning, with Python and some C++ for optimisation helpful. NCP-AAI asks for one to two years in AI/ML roles plus hands-on production agentic work. NCA-GENL asks only for a basic understanding of generative AI and large language models, and its five areas are Core Machine Learning and AI Knowledge (30%), Software Development (24%), Experimentation (22%), Data Analysis (14%) and Trustworthy AI (10%).

How to choose

Ask what you spend your week changing. If it is weights, kernels, quantisation, batch sizes and fine-tuning runs, NCP-GENL describes your job. If it is prompts, tools, orchestration, retrieval, evaluation harnesses and guardrails around a model someone else serves, NCP-AAI does. If you are still learning what either sentence means, start with NCA-GENL: it is $125, an hour long and conceptual, and it will tell you which half of the field you want to go deeper in. All three are valid for two years, and NVIDIA recertifies by retaking the exam.