NCA-MMGA Multimodal GenAI Associate Study Plan¶

6-Week Study Schedule¶

Week 1: Multimodal AI Fundamentals¶

• Study modality types (text, image, audio, video)
• Learn VLM architecture (image encoder + LLM)
• Understand CLIP and cross-modal alignment
• Study fusion strategies (early, late, cross-attention)
• Reference: NVIDIA Build - Try multimodal models

Week 2: Image and Video Generation¶

• Study diffusion model basics (forward/reverse process)
• Learn text-to-image prompting and control parameters
• Understand image-to-image, inpainting, outpainting
• Study video generation concepts
• Reference: NVIDIA Developer Blog

Week 3: Speech and Audio¶

• Study ASR pipeline and Word Error Rate
• Learn TTS pipeline and naturalness concepts
• Understand NVIDIA Riva architecture
• Study audio processing fundamentals
• Reference: Riva Documentation

Week 4: NVIDIA Tools and Applications¶

• Study NIM for multimodal model serving
• Learn NeMo for multimodal training
• Understand multimodal RAG architecture
• Study content moderation across modalities

Week 5: Review and Practice¶

• Work through practice scenarios
• Review all domains
• Create flashcards for key concepts
• Focus on weak areas

Week 6: Final Preparation¶

• Timed practice session
• Review fact sheet
• Light review and exam logistics

Self-Assessment¶

• Can I explain how diffusion models work?
• Do I understand VLM architecture?
• Can I describe NVIDIA Riva capabilities?
• Do I know multimodal RAG architecture?
• Can I explain evaluation metrics (FID, CLIP Score)?