Free Practice Questions for IBM watsonx Generative AI Engineer v1 - Associate Certification

Subdomain 1.1: Understand the 5 Capabilities of GenAI models/LLMs

• Summarization
• Classification
• Generation (Code, Translation)
• Extraction
• Q&A

Subdomain 1.2: Articulate the Components in Gen AI Patterns

• Mixture of experts (MoE)
• Variation Auto Encoders (VaE)
• Transformer based models
• Reasoning models

Subdomain 1.3: Understand the Limitations of GenAI/LLMs

• Technical limitations
• Ethical limitations
• Bias mitigation
• Risk evaluation

Subdomain 1.4: Understand Use Case and Identify Gen AI Application Opportunities

• Industry use cases
• Needs analysis
• Solution proposal
• Feasibility reports

Subdomain 1.5: Understand How to Choose the Appropriate Model for a Use Case

• Model selection criteria
• Parameter size
• Chat vs instruct
• IBM Granite models
• Billing classes

Subdomain 1.6: Articulate the Optimal Model Architecture Based on Use Case

• Model architecture definition
• Component examination
• Agentic architectures

Subdomain 1.7: Identify and apply various tools and techniques like AI agents, RAG, LangChain, etc.

• RAG Pattern
• LangChain
• AI agents

Subdomain 1.8: Understand security risks associated with LLMs, prompt engineering, prompt, and data

• Data Bias
• Data Poisoning
• Data Curation
• Data Privacy
• Prompt Injection
• Prompt Leaking
• Output Bias
• Toxic Output
• Harmful code generation
• Guardian models

Subdomain 2.1: Differentiate between zero-shot and few-shot prompting

• Zero-shot prompting
• Few-shot prompting

Subdomain 2.2: Design Prompts based on use case

• Model selection
• Chat use case
• Translation use case

Subdomain 2.3: Generate Prompt Templates

• Prompt templates
• Evaluation
• Creation
• Deployment
• Tracking

Subdomain 2.4: Determine the best model parameters for each GenAI prompt

• Decoding
• Greedy Decoding
• Sampling Decoding
• Random Seed
• Repetition Penalty
• Stopping Criteria
• Stop Sequences
• Minimum Tokens
• Maximum Tokens

Subdomain 2.5: Describe the benefits of using prompt variables

• Prompt variables
• Reusable prompts

Subdomain 2.6: Describe the benefits of Prompt Lab

• Prompt editing (Chat, Structured, Freeform)
• Reusable prompts
• Input prompts

Subdomain 2.7: Controlling model parameters

• Greedy decoding
• Sampling decoding
• Temperature
• Top K
• Top P
• Random Seed
• Repetition penalty
• Stop sequences
• Minimum tokens
• Maximum tokens
• Model generation time limit

Subdomain 2.8: Articulate model risks

• Hallucinations
• Personal information risks
• PII filter
• Hate speech
• HAP filter
• Bias
• Data bias
• Data poisoning

Subdomain 3.1: Understand the difference between hard and soft prompts

• Hard prompts
• Soft prompts
• Readability
• Explainability
• Performance
• Simplicity
• Interpretability

Subdomain 3.2: Reconstruct prompts to reduce the cost of using GenAI models

• Token usage management
• Inefficient prompt detection
• Cost-effective templates
• Model parameters (Stop sequences, Min/max token limits)

Subdomain 3.3: Plan for Data elements for application usage

• watsonx.ai project data
• Data Refinery

Subdomain 3.4: Articulate model quantization techniques

• LLM quantization
• Precision reduction
• Computational cost reduction

Subdomain 3.5: LoRA

• LoRA

Subdomain 3.6: Prepare the dataset for training

• Taxonomy tree-based curation
• Synthetic data generation
• InstructLab
• LAB methodology

Subdomain 3.7: Customize LLMs with InstructLab

• InstructLab components
• Taxonomy driven curation
• Synthetic data generation
• Alignment tuning (Knowledge, Skill)
• InstructLab workflow

Subdomain 3.8: Generate synthetic data using the User Interface

• Existing data leverage
• Custom data schema
• Data import limitations
• Anonymization
• Kolmogorov-Smirnov
• Anderson-Darling
• Differential privacy
• Privacy budget
• Privacy leakage probability
• Random seed
• Sizing requirements

Subdomain 4.1: Describe what embeddings are in Context of GenAI

• Text embeddings
• IBM Embedding models
• Third party embedding models

Subdomain 4.2: Generate vector embeddings utilizing models

• Text to embedding vectors
• Embedding API prerequisites
• Vector databases (Purpose built, Extensions)

Subdomain 4.3: Describe when to use a vector database

• Retriever concept
• Vector databases (Embedded, Static)
• Watson Discovery
• GitHub code retrieval API
• watsonx Discovery
• Retriever capabilities
• Use case driven selection

Subdomain 4.4: Develop using libraries and tools

• RAG Pattern
• LangChain
• WatsonX LLM
• Watson ML
• ElasticSearch
• SingleStore
• LlamaIndex
• Chunking/text splitting
• Agentic RAG
• AutoRAG

Subdomain 5.1: Plan a deployment based on client needs

• Prompt template lifecycle
• Prompt template changes
• AI Asset roles
• AI Governance
• Model performance
• Inference evaluation
• Prompt outcome explanation

Subdomain 5.2: Deploy AI Assets

• Deployment space benefits
• Prompt template deployment
• Application changes

Subdomain 5.3: Deploy a custom model

• watsonx requirements
• Foundation model
• Custom model access

Subdomain 5.4: High-level architecture for deployment options

• Deployment spaces versioning
• Prompt version changes
• Prompt template testing
• Model gateway

Subdomain 5.5: Plan the deployment of prompts for versioning

• SaaS vs Software
• Architectural patterns (RAG, Summarization, Q&A)
• Data repository (AI Pipelines, Data management, Chunk size)
• Endpoint security
• PromptLab code generation

Subdomain 6.1: Integrate watsonx.ai with Other Services

• IBM Cloud services integration (Watson Assistant, Watson Discovery, watsonx.governance)
• Integration points
• Data exchange
• UI/API configuration
• watsonx.ai APIs/SDKs
• LangChain

Subdomain 6.2: Orchestrate AI Workflows

• AI workflow design
• LangChain-based chains
• Orchestration tools
• Scheduling
• Dependency management
• Error handling
• Conditional logic
• Workflow monitoring
• Agentic RAG

Subdomain 6.3: Understand real-world Integration Scenarios

• Business use case solution design
• Data sources
• IBM Cloud services
• External systems
• High-level architecture
• Implementation
• Testing
• Validation
• Model Context Protocol

Subdomain 6.4: Develop LLM based applications with LangChain

• LangChain core concepts (chains, agents, tools, memory)
• Conversational AI
• Generative AI applications
• Modularity
• Reusability
• Extensibility
• LangChain chains
• LLMs
• Prompt templates
• External data sources
• LangChain agents