CAIM study plan

Draft a one-page AI strategy canvas that states business objectives, value hypotheses, and the governance decision rights for a fictional or real organization.

Classify at least 10 AI use cases by strategic fit and risk tier, and justify each placement in two sentences.

Map the AI governance operating model: who approves, who monitors, who escalates, and which artifacts record each decision.

Write a glossary of 15 core governance terms (model owner, risk appetite, human oversight, accountability) with a one-line definition each.

Build a risk register for the use cases from week 1 with likelihood, impact, mitigation owner, and residual risk rating.

Compare at least three regulatory or standards frameworks relevant to AI (for example NIST AI RMF, ISO/IEC 42001, and the EU AI Act) on scope, obligations, and evidence required.

Define an ethics review gate with explicit pass/fail criteria for bias testing, transparency, and human oversight.

Identify which use cases would be classified high-risk under a chosen regime and state the added controls each needs.

Construct a cost-benefit model for one AI initiative that includes build, run, data, and change-management costs over three years.

Quantify at least three benefit categories (revenue, cost avoidance, cycle time) and label each assumption as measured, benchmarked, or estimated.

Calculate NPV, payback period, and a sensitivity case for a 30% adoption shortfall.

Draft a value-realization scorecard that links each KPI to a baseline, a target, and a measurement cadence.

Design a data architecture diagram covering ingestion, storage, quality checks, lineage, and access control for one AI workload.

Write data-quality acceptance criteria for at least five dimensions (completeness, accuracy, timeliness, consistency, representativeness) with measurable thresholds.

Evaluate compute options (cloud GPU, managed AI service, on-premises) against latency, cost, and data-residency constraints, and recommend one.

Document a data-retention and privacy-handling policy for training and inference data, including deletion paths.

Sequence a full AI solution lifecycle from problem framing through decommissioning, naming the exit criteria for each stage.

Write a model evaluation plan specifying offline metrics, an online pilot design, and acceptance thresholds agreed with the business owner.

Produce a delivery risk log with dependencies, owners, and mitigation dates for a 12-week pilot.

Define the go/no-go review agenda and the evidence pack required before production release.

Create a role-and-skills map for the AI team covering data engineering, ML engineering, product ownership, risk, and domain expertise, with build-versus-buy decisions.

Draft a 90-day change-management plan with stakeholder segments, communication cadence, training formats, and adoption metrics.

Write an incident-response runbook for an AI failure, including detection, containment, notification, and post-incident review.

Prepare a 10-slide executive briefing that translates technical risk into business language.

Define a monitoring dashboard with at least 8 signals spanning model performance, data drift, cost, latency, and safety incidents.

Set alert thresholds and an escalation path for each signal, and state the owner and response time.

Design a continuous-improvement loop with a retraining trigger policy and a rollback plan.

Run a tabletop exercise on a drift alert and record the decisions and gaps found.

Complete two timed 60-question practice sets and score at least 80% on each, logging every missed item by domain.

Re-study the two weakest domains using the logged misses and rewrite each missed concept in your own words.

Recite from memory the governance decision rights, the risk-tier definitions, and the lifecycle stage gates.

Assemble a one-page exam-day reference sheet of formulas, frameworks, and thresholds.