HIRAF study plan

Define core AI vocabulary (agent, model, inference, training) in your own words and map each term to a real example

Distinguish AI, machine learning, deep learning, and generative AI by writing a one-page comparison with boundaries

Classify ten everyday AI applications by technique type and explain the reasoning for each classification

Summarize the standard machine learning workflow from problem framing through deployment and monitoring

Describe supervised, unsupervised, and reinforcement learning and match each to two concrete use cases

Explain how training data, features, labels, and loss functions interact during model fitting

Identify sources of bias and variance and predict how each affects train versus test performance

Compare at least four evaluation metrics and state when each is misleading

Walk through the full model development lifecycle and list the decision gate at each stage

Explain overfitting and underfitting and describe three mitigation techniques for each

Outline how validation, cross-validation, and test sets prevent evaluation leakage

Draft a model card template capturing intended use, training data, metrics, and limitations

Compare data quality dimensions (accuracy, completeness, consistency, timeliness) and describe how each degrades model performance

Explain data lineage, provenance, and metadata and why each matters for reproducibility

Identify personal and sensitive data categories and describe anonymization and pseudonymization approaches

Map the data lifecycle from collection through retention and deletion with the control at each stage

Define the core AI risk categories (performance, security, privacy, fairness, transparency, misuse) and give an example of each

Apply a risk assessment method to rank at least five risks by likelihood and impact

Explain fairness metrics and describe how a chosen metric encodes a value judgment

Distinguish explainability from interpretability and select appropriate explanation methods for three model types

Summarize the purpose and structure of major AI governance frameworks and principle sets at a conceptual level

Explain how regulation assigns obligations across developers, deployers, and users of AI systems

Describe what a conformity assessment, audit, and impact assessment each require and when each applies

Connect ethical principles such as accountability and human oversight to concrete organizational controls

Describe how to operationalize governance through policies, roles, approval gates, and documentation

Explain monitoring, drift detection, and incident response for deployed AI systems

Outline logging, versioning, and rollback requirements for reproducible production AI

Identify security threats specific to AI systems, including data poisoning and prompt injection, and describe one control for each

Consolidate all prior deliverables into a single indexed study portfolio

Complete a timed 60-question practice set and analyze every miss by domain and root cause

Re-teach the three weakest domain areas using your own summaries without notes

Build a one-page formula and framework reference sheet for final review