OCI AI Foundations Study Plan¶

Three weeks at 4-5 hours per week. Shorter than most plans in this repo, because the exam is 40 questions at 65%.

Week 1: AI and machine learning fundamentals¶

• Enrol in the free Oracle University learning path
• AI, machine learning, and deep learning: how they nest
• Task categories: language, speech, vision, decision
• Supervised learning: classification versus regression, with examples of each
• Unsupervised learning: clustering, dimensionality reduction, anomaly detection
• Reinforcement learning at a conceptual level
• The ML workflow end to end
• Train, validation, and test splits, and why the test set is held back
• Overfitting versus underfitting, and the bias-variance trade-off
• Evaluation metrics: accuracy, precision, recall, F1, confusion matrix
• Review Notes: notes/01-ai-and-ml-fundamentals.md

Week 2: Deep learning and generative AI¶

• Neural network anatomy: neurons, layers, weights, activation functions
• Training: loss, backpropagation, gradient descent
• Hyperparameters: learning rate, epochs, batch size
• CNNs for vision, RNNs and LSTMs for sequences, transformers for language
• Why GPUs are used for training
• Generative versus discriminative models
• Transformers, attention, tokens, embeddings
• Pre-training, fine-tuning, instruction tuning
• Prompt engineering: zero-shot, few-shot, chain-of-thought
• Context window, hallucination, and grounding
• RAG and vector search
• Review Notes: notes/02-deep-learning.md and notes/03-generative-ai-and-llms.md

Week 3: OCI services and review¶

• OCI Generative AI and Generative AI Agents
• OCI Language, Speech, Vision, Document Understanding
• OCI Data Science: notebooks, model catalog, deployments, jobs
• OCI Data Labeling
• Select AI and AI Vector Search in Autonomous Database
• AI infrastructure: GPU shapes, bare metal, RDMA cluster networking
• Drill the service selection table until it is automatic
• Responsible AI principles
• Review Notes: notes/04-oci-ai-services.md
• Two timed practice exams

Readiness check¶

• Explain how AI, machine learning, and deep learning relate
• Give an example of classification, regression, and clustering
• Explain overfitting and one way to reduce it
• Explain when accuracy is a misleading metric, and what to use instead
• Name the OCI service for sentiment, transcription, object detection, and invoice extraction
• Explain what RAG does and why it reduces hallucination
• Explain what a context window is
• Say which OCI service you would use to fine-tune a foundation model