Pearson Vue Certified IT Specialist Artificial Intelligence Training

This course prepares you for the INF-307 certification exam, covering all official exam domains:

Domain 1 AI Problem Definition

• Identify the problem to solve using AI (need, inputs/outputs, whether AI is warranted, measurable success)
• Classify the problem type (e.g., classification, regression, unsupervised, reinforcement) based on available data
• Identify areas of expertise needed (business, domain, AI, implementation)
• Build a security plan (access levels, infrastructure security, attack-surface/risk assessment)
• Ensure AI is used appropriately (guard against misprediction/harm, set data and algorithm-selection guidelines)
• Choose transparency and validation activities (communicate data-collection purpose, review legal/industry requirements)

Domain 2 Data Collection, Processing, and Engineering

• Choose data collection methods and determine whether an existing dataset can be used or a new one must be generated
• Assess data quality and ensure data are representative (check for missing/corrupt data and bias)
• Identify resource requirements (compute, time complexity, budget)
• Convert data into suitable formats and select/engineer features for the AI model
• Identify training and test datasets
• Document data decisions for transparency to regulators and end users

Domain 3 AI Algorithms and Models

• Evaluate applicability of specific algorithm families (e.g., neural network, decision tree, k-means) to the problem
• Train and tune a model, gathering performance metrics and iterating
• Select a specific model after experimentation while avoiding overengineering, considering cost/speed/explainability
• Evaluate model performance (accuracy, precision, overfitting/underfitting, cross-validation) and sensitivity/specificity
• Look for potential sources of bias in the algorithm and training data
• Confirm regulatory adherence and obtain stakeholder approval

Domain 4 Application Integration and Deployment

• Train customers on product use, limitations, and expectations; share documentation
• Plan to address potential challenges of models in production
• Design a production pipeline (training/prediction) and integrate the AI with the application
• Test accuracy, robustness, and speed of the AI through the application
• Support the AI solution (maintenance documentation, support team, feedback mechanism, drift detector)

Domain 5 Maintaining and Monitoring AI in Production

• Engage in oversight: log performance, monitor systems, act on alerts, watch for drift/degraded operation
• Assess business impact via key performance indicators, comparing pre/post-change metrics
• Measure impacts on individuals and communities; identify and mitigate subgroup issues
• Handle user feedback (satisfaction, confusion) and incorporate it into future versions
• Decide on a regular basis whether to retrain, continue using as-is, or decommission the AI