AWS has published the full DVA-C03 exam guide for AWS Certified Developer - Associate. In September we only had domain names, weights and a list of themes. Now there are real task statements, a skill-by-skill comparison with DVA-C02, and updated in-scope and out-of-scope service lists.
DVA-C03 domains and official weights
| Domain | DVA-C02 | DVA-C03 |
|---|---|---|
| 1. Development with AWS Services | 32% | 30% |
| 2. Security | 26% | 26% |
| 3. Testing and Deployment (C02: Deployment) | 24% | 22% |
| 4. Troubleshooting and Optimization | 18% | 22% |
The weights match what AWS shared in September. Development drops two points, Deployment becomes Testing and Deployment and also drops two, and Troubleshooting and Optimization gains four. Security stays at 26% but gets a whole new task.
What is new in the guide
- Task 2.3, security of AI services. Requesting and managing access to AI services, sending data with privacy controls (VPC endpoints, keeping inputs and outputs out of model training), filtering model inputs and outputs (content filters, PII and PHI redaction, denied topics, prompt injection protection), securing AI agent interactions (tool-use authorization, session isolation, human-in-the-loop approval), and keeping sensitive content out of monitoring logs.
- AI-assisted development across the domains. Generating, reviewing and optimizing code and integrating managed AWS AI services (Task 1.1), generating tests and automating regression testing (Task 3.2), supporting CI/CD workflows (Task 3.3), analyzing errors (Task 4.1) and finding optimization opportunities (Task 4.3).
- Containers. Building and managing images with Amazon ECR (Dockerfiles, tagging, lifecycle policies) in Task 3.1, and deploying with ECS task definitions, EKS manifests, AWS Fargate and rolling updates in Task 3.3.
- Presigned URLs. Time-limited access with Amazon S3 and Amazon CloudFront is now spelled out in Task 2.1.
New in-scope names include Amazon Bedrock, Amazon Bedrock AgentCore, Amazon Quick, Kiro, Amazon Data Firehose and AWS PrivateLink. Amazon SageMaker AI and Strands Agents are out of scope.
What moved or got merged
- Unit testing with AWS SAM moved from Task 1.1 to Task 3.2.
- Lambda performance tuning moved from Task 1.2 to Task 4.3.
- Serializing and deserializing data moved from Task 1.3 to Task 1.1.
- API Gateway stages and custom domains moved from Task 3.4 to Task 3.1.
AWS also merged many small skills. The four DynamoDB skills are now one, encryption and sensitive data went from 13 skills over two tasks to 7 in one task, and testing went from 11 skills to 5. AWS says this does not reduce emphasis.
Exam format
65 questions (50 scored, 15 unscored), 130 minutes, a scaled score of 100 to 1,000 with 720 to pass, and compensatory scoring. Question types are multiple choice and multiple response, delivered through Pearson VUE at a test center or online.
The target candidate has one or more years of hands-on experience building and maintaining applications on AWS. You are not expected to design architectures, design CI/CD pipelines, administer IAM users and groups, select foundation models, write prompts, design RAG architectures or design AI guardrails and governance.
Key dates
- 27 Oct 2026: DVA-C03 registration opened and the updated guide was released
- 30 Nov 2026: last day to take DVA-C02
- 1 Dec 2026: DVA-C03 delivery begins
If you are ready, take DVA-C02 by 30 November; your certification stays active until its normal expiry date. If you are early in your prep, or already use AI coding tools and Bedrock at work, C03 is the better fit. Book early either way.
How to study from here
- Read the official guide, task statements first, then the comparison page and the in-scope service list.
- Keep the classics. Lambda, DynamoDB, SQS, SNS, EventBridge, API Gateway, IAM roles, KMS and CodeDeploy still make up most of the task statements.
- Give Troubleshooting and Optimization more time. Practice CloudWatch logs and metrics, X-Ray traces, and Lambda concurrency and memory issues.
- Build one small app that touches the new tasks, for example a Lambda function that calls Bedrock through a VPC endpoint, applies a content filter, keeps prompts out of the logs, ships as a container image in ECR and deploys to ECS on Fargate with a rolling update.
- Skip what is out of scope: no prompt engineering, long RAG design courses or SageMaker AI notebooks.
Want to check where you stand? Try the free sample exam quiz - 5 questions, no signup.
Originally published on testimea
Top comments (0)