Study guide
1.2 AIP-C01 Scoring & Professional Decision Strategy
AWS Certified Generative AI Developer - Professional AIP-C01
Track your progress as you read
Finished sections get a check mark.
1 Chapter 1: Exam Overview & Foundation Model Fundamentals
1.1 AIP-C01 Exam Structure, Logistics & Passing Strategy 1.2 AIP-C01 Scoring & Professional Decision Strategy 1.3 Foundation Model Landscape & Model Selection on Amazon Bedrock 1.4 Model Abstraction, Controlled Switching & Resilience 1.5 Customization Spectrum: Prompting vs RAG vs Fine-Tuning vs Continued Pre-training 1.6 Custom Model Deployment & Lifecycle Management
2 Chapter 2: Bedrock Knowledge Bases & RAG Architectures
2.1 Managed RAG Architecture with Bedrock Knowledge Bases 2.2 Knowledge Source Integration, Synchronization & Freshness 2.3 Data Ingestion, Parsing & Chunking Strategies 2.4 Embedding Models & Vector Representations 2.5 Vector Index Scale, Migration & Retrieval Evaluation 2.6 Vector Database Selection: OpenSearch Serverless vs Aurora pgvector vs Managed Stores 2.7 Vector Store Security, Multi-Tenancy & Connectivity
3 Chapter 3: Advanced Vector Search, Indexing & Data Compliance
3.1 Hybrid Search, Metadata Filtering & Semantic Reranking 3.2 Metadata Filtering, Reranking & Query Transformation 3.3 Fine-Tuning & Continued Pre-Training Data Pipeline Engineering 3.4 Data Validation, Multimodal Processing & Model Formatting 3.5 Data Privacy, Data Residency & Compliance Governance 3.6 Privacy Preservation, Source Lineage & Audit Evidence
4 Chapter 4: Prompt Engineering & Bedrock Runtime APIs
4.1 Advanced Prompt Engineering Techniques & Inference Parameters 4.2 Inference Parameters, Prompt Governance & Regression 4.3 Converse API vs InvokeModel API & Message Formatting 4.4 Tool Calling, FM API Resilience & Model Routing 4.5 Streaming Architectures & Asynchronous Event Processing 4.6 Resilient Streaming, Deployment & Asynchronous Processing
5 Chapter 5: Amazon Bedrock Agents & Tool Integration
5.1 Agentic Architecture, ReAct Reasoning Loops & Orchestration 5.2 Agent Frameworks, MCP & Lifecycle Management 5.3 Action Groups, OpenAPI Schemas & AWS Lambda Integration 5.4 Agent Memory, Session State & Multi-Agent Collaboration Patterns 5.5 Persistent State & Multi-Agent Coordination 5.6 Advanced Agent Prompt Templates & Flow Customization 5.7 Agent Safeguards, Tool Reliability & Cost Control
6 Chapter 6: Workflow Orchestration & Application Integration
6.1 Amazon Bedrock Prompt Flows & Visual Orchestration 6.2 Step Functions Orchestration & Human-in-the-Loop Workflows 6.3 Enterprise Integration: API Gateway, AppSync, and EventBridge 6.4 Hybrid Integration, GenAI Gateways & CI/CD Tooling
7 Chapter 7: AI Safety & Guardrails for Amazon Bedrock
7.1 Guardrails Architecture, Denied Topics & Content Filters 7.2 Content Safety, Defense in Depth & Threat Detection 7.3 Sensitive Information Filters & PII Redaction/Masking 7.4 Contextual Grounding Checks & Hallucination Prevention
8 Chapter 8: Security, Access Control & Network Architecture
8.1 IAM Least Privilege & Resource Policies for Bedrock 8.2 PrivateLink VPC Endpoints & Network Isolation 8.3 Encryption with AWS KMS & Cross-Account Access Patterns
9 Chapter 9: Governance, Auditing & Operational Logging
9.1 SageMaker Model Cards & AI Governance Frameworks 9.2 Lineage, Fairness, Transparency & Continuous Governance 9.3 Invocation Logging, CloudTrail Auditing & Observability 9.4 Responsible AI, Bias Mitigation & Red Teaming Methodologies
10 Chapter 10: Performance, Throughput & Cost Optimization
10.1 On-Demand vs Provisioned Throughput & Capacity Planning 10.2 Capacity Measurement, Scaling Patterns & Resilience 10.3 Batch Inference & Bedrock Intelligent Prompt Routing 10.4 Intelligent Prompt Routing & Application Eligibility 10.5 Prompt Caching, Token Optimization & Latency Reduction
11 Chapter 11: Testing, Evaluation & Troubleshooting GenAI Solutions
11.1 Amazon Bedrock Model Evaluation (Automatic & Human Workflows) 11.2 Human, RAG & Continuous Evaluation 11.3 Evaluation Metrics & RAG Quality Benchmarking (Ragas, ROUGE, BLEU) 11.4 RAG and Agent Performance Evaluation 11.5 CloudWatch Metrics, Throttling & Operational Observability 11.6 Quota Diagnosis, Retry Control & Operational Troubleshooting 11.7 Troubleshooting Common GenAI Failure Modes & Agent Traces 11.8 Prompt Maintenance, Content Handling & Deployment Validation
1.2 AIP-C01 Scoring & Professional Decision Strategy
Key Takeaways Explain This Topic AI
AIP-C01 has 65 scored and 10 unscored questions in a 180-minute appointment.
Multiple-response items require every requested correct response; the item states how many to select.
Constraint words such as least overhead or lowest latency must be evaluated against every stated requirement.
Last updated: September 2026
- 1.2 AIP-C01 Scoring & Professional Decision Strategy
- Scoring Mechanics & Question Formats
- Compensatory Scoring
The AIP-C01 exam uses a compensatory scoring model. This means there are no domain-specific minimum passing scores; a candidate who achieves a lower score in Domain 4 or Domain 5 can still pass the examination provided their high performance in Domains 1, 2, and 3 elevates their total scaled score to 750 or above. Furthermore, there is no penalty for incorrect answers. Unanswered questions and incorrect selections receive zero credit, making it critical that candidates never leave any question unanswered.
Question Formats and the Zero Partial Credit Rule
The exam consists of two question styles:
Multiple-choice: Exactly one correct option out of four choices.
Multiple-response: Two or more correct responses from five or more options; the item tells you how many responses to select.
Important
AWS does not award partial credit on multiple-response questions. If a question prompts you to "Select TWO" and you select one correct option and one incorrect option, you receive zero points for the entire item. Every choice must be individually verified against the scenario requirements.
Unscored Pretest Items
Out of the 75 questions presented, exactly 10 are unscored pretest items used by AWS to statistically evaluate candidate performance on newly authored questions before introducing them into future scored banks. These 10 items are indistinguishable from the 65 scored items and are distributed randomly throughout the exam. Candidates must treat every question with identical focus and avoid attempting to guess whether a question is pretest or scored.
Pacing Strategy & Time Management Architecture
With 180 minutes to complete 75 questions, candidates have an average of 2.4 minutes per item. Professional-level scenario questions typically feature 2 to 4 paragraphs of technical requirements, architectural constraints, and four verbose response options. Adopting an unstructured approach risks running out of time on later questions.
A proven three-pass execution strategy provides structured time management: Pass 1: Rapid Sweep & Baseline Answers (Minutes 0 – 120)
- Answer high-confidence questions within 60–90 seconds.
- For complex scenario items, eliminate obvious distractors, record a provisional answer, and flag the question.
- Target: Complete initial pass across all 75 questions.
Pass 2: Flagged Item Deep Analysis (Minutes 120 – 160)
- Focus exclusively on the 15–20 flagged questions.
- Allocate 2 to 2.5 minutes per flagged question.
- Re-read constraints, map qualifiers, and verify multi-response selections.
Pass 3: Final Verification & Sanity Check (Minutes 160 – 180)
- Verify that all 75 questions have an answer recorded.
- Confirm no accidental misclicks occurred.
- Ensure zero unanswered items remain.
Deconstructing Professional Question Stems & Constraint Qualifiers
Success on the AIP-C01 exam requires deciphering constraint qualifiers. When analyzing a question stem, identify the exact discriminator phrase that defines the winning architecture:
"LEAST operational overhead":
Architecture Meaning: Eliminate custom infrastructure, self-hosted containers, manual scaling configurations, and custom cron synchronization tasks.
AWS Selection: Choose native, fully managed AWS features. For example, select Amazon Bedrock Knowledge Bases with managed OpenSearch Serverless rather than provisioning a self-managed OpenSearch cluster on Amazon EC2 or configuring custom vector indexing Lambda workers.
"MOST cost-effective":
Architecture Meaning: Minimize idle compute, compute hours, and token consumption.
AWS Selection: Use smaller, specialized models (such as Amazon Nova Micro or Claude 3 Haiku) for classification and routing instead of frontier models. Select Bedrock Batch Inference for asynchronous document processing (which offers a 50% discount compared to on-demand pricing). Avoid Provisioned Throughput unless continuous, high-volume 24/7 utilization justifies fixed hourly Model Units.
"LOWEST latency":
Architecture Meaning: Minimize time-to-first-token (TTFT) and overall generation duration.
AWS Selection: Leverage streaming APIs ( InvokeModelWithResponseStream or ConverseStream ), prompt caching for static prefixes, Provisioned Throughput to eliminate queuing bottlenecks, and smaller model architectures.
"STRICTEST security / regulatory compliance":
Architecture Meaning: Prevent data exposure, enforce encryption at rest and in transit, restrict public network egress, and redact sensitive data.
AWS Selection: Configure AWS KMS customer-managed keys (CMK), interface VPC endpoints (AWS PrivateLink), Amazon Bedrock Guardrails with PII masking and contextual grounding checks, and ensure Bedrock model invocation logging to CloudWatch Logs or S3 is encrypted and access-restricted.
Common Exam Traps to Avoid
The Over-Engineering Trap: Selecting custom model fine-tuning or continued pre-training when prompt engineering, few-shot examples, or Bedrock Knowledge Bases fully solve the problem with a fraction of the cost and complexity.
The Control Plane vs. Data Plane Trap: Confusing administrative IAM actions with inference actions. For example, generating model responses requires bedrock:InvokeModel or bedrock:Converse , whereas listing models requires bedrock:ListFoundationModels .
The Incomplete Verification Trap on Multi-Response Questions: Confirming only one valid architectural component and guessing the second. Both selections must be fully validated against the scenario constraints.
The Premature Model Escalation Trap: Selecting the most capable model in the catalog (such as Claude 3 Opus or Amazon Nova Pro) for simple categorization or sentiment analysis tasks, violating cost and latency efficiency principles.