NVIDIA Certified Professional - Generative AI LLMs

Validates professional-level expertise in developing, optimizing, and deploying large language models using NVIDIA platforms, including LLM architecture and transformer internals, prompt engineering with chain-of-thought and few-shot techniques, data preparation and tokenization, model optimization with quantization and pruning, fine-tuning with SFT and PEFT/LoRA, evaluation methodology, GPU acceleration and optimization for distributed training, model deployment with TensorRT-LLM and Triton, production monitoring and reliability, and safety and compliance. The exam covers ten domains: Model Optimization (17%), GPU Acceleration and Optimization (14%), Prompt Engineering (13%), Fine-Tuning (13%), Data Preparation (9%), Model Deployment (9%), Evaluation (7%), Production Monitoring and Reliability (7%), LLM Architecture (6%), and Safety, Ethics, and Compliance (5%). Format: 60-70 multiple-choice questions, 120 minutes, proctored online.