NCP-AAI is NVIDIA's professional-level agentic AI exam. It runs for 120 minutes with 60 to 70 questions, costs $200 and is valid for two years. It is English only and remotely proctored. NVIDIA recommends one to two years in AI/ML roles plus hands-on work on production agentic projects. The blueprint has ten domains: agent architecture and design (15%), agent development (15%), evaluation and tuning (13%), deployment and scaling (13%), cognition, planning and memory (10%), knowledge integration and data handling (10%), NVIDIA platform implementation (7%), run, monitor and maintain (5%), safety, ethics and compliance (5%), and human-AI interaction and oversight (5%). That weighting is the best thing about it.
Most agent courses spend their time on wiring up tools and a loop. This one puts as much weight on evaluating and deploying agents as on building them, and only 7% is about NVIDIA's own products. That makes it more vendor-neutral than I expected, and closer to what an engineer actually needs to know. My complaints are about cost and depth in the places that matter. The four recommended prep courses on the Deep Learning Institute range from about $30 to $500 each, so a candidate who follows the official path can spend several times the exam fee.
And monitoring, safety and oversight together are only 15%. Anyone who has run an agent in production knows those are where the incidents happen. Search interest jumped in late 2025 when the exam launched and has settled since, but it remains one of the more credible agent certifications. If you already build agents at work, it is worth sitting. If you do not, start with free material like Hugging Face's agents course and come back once you have something in production.