AWS Certified Developer – Associate is changing. The new version, DVA-C03, is used from 1 December 2026. The last day to take the old DVA-C02 is 30 November 2026. Registration for DVA-C03 opened on 27 October 2026. The biggest change is AI: DVA-C03 tests AI-assisted development tools, integrating managed AWS AI services, and a new task on the security risks of using AI services in your apps.
This guide follows the official AWS exam guide: the four domains with weights, every task, what changed from C02, a 6-week plan and 15 original scenario questions.
Quick facts (AWS, November 2026)
| Item | Detail |
|---|---|
| Exam code | DVA-C03 (from 1 Dec 2026); DVA-C02 until 30 Nov 2026 |
| Length | 130 minutes |
| Questions | 65 (50 scored + 15 unscored), multiple choice or multiple response |
| Passing score | 720 on a 100–1,000 scale (compensatory: you pass on the overall score) |
| Fee | USD 150 |
| Languages | English, Japanese, Korean, Portuguese (Brazil), Simplified Chinese, Spanish (Latin America) |
| Delivery | Pearson VUE test centre or online proctored |
| Valid for | 3 years |
| Recommended experience | 1+ year developing and maintaining applications on AWS |
Check the official Developer – Associate page and the DVA-C03 exam guide before booking.
Should you take DVA-C02 before 30 November or wait for DVA-C03?
- Take C02 only if you have already prepared with C02 material and can sit the exam before 30 November 2026.
- Prepare for C03 if you are starting now. Both earn the same certification, and C03 matches what employers now expect: AI coding assistants, containers and AI service security.
Who should take it?
- Developers who build on Lambda, API Gateway, DynamoDB, SQS and S3.
- Freshers who know one programming language well and want a hands-on AWS credential after Cloud Practitioner.
- Engineers heading towards DevOps Engineer – Professional, which also renews this certification.
Domains and weights
| Domain | DVA-C02 | DVA-C03 |
|---|---|---|
| 1. Development with AWS Services | 32% | 30% |
| 2. Security | 26% | 26% |
| 3. Testing and Deployment (was "Deployment") | 24% | 22% |
| 4. Troubleshooting and Optimization | 18% | 22% |
What's new in DVA-C03
- New Task 2.3: security risks of AI services: requesting access to AI services, private connectivity (VPC endpoints), making sure inputs and outputs aren't used for training, content filtering, PII/PHI redaction, prompt-injection protection, securing AI agents (tool-use authorisation, session isolation, human approval) and keeping sensitive content out of logs.
- AI-assisted development: generating, reviewing and optimising code with AI tools; AI-generated tests; AI support in CI/CD; AI troubleshooting and optimisation suggestions.
- Containers: building images with ECR, Dockerfiles, tagging and lifecycle policies; deploying with ECS task definitions, EKS manifests and Fargate.
- Presigned URLs for S3 and CloudFront.
- Moved: SAM unit testing to Domain 3; Lambda performance tuning to Domain 4. DynamoDB is now one combined skill.
Out of scope: designing architectures, designing CI/CD pipelines, selecting foundation models, prompt engineering and RAG design. Those belong to Solutions Architect and the AI exams.
Domain 1: Development with AWS Services (30%)
- Application code: loose coupling; synchronous (API Gateway) vs asynchronous (SQS, EventBridge); retries with backoff and graceful degradation; API request/response transformations and validation; SDKs; streaming data; serialisation; event-driven patterns with EventBridge, S3 Event Notifications, SQS and SNS; AI-assisted coding; integrating managed AI services.
- Lambda: VPC access to private resources; memory, concurrency, timeout, layers and extensions; destinations and dead-letter queues; event source mappings.
- Data stores: lifecycles (S3 Lifecycle, TTL); choosing ElastiCache, DynamoDB, S3 or RDS by access pattern; caching; DynamoDB keys, indexes, consistency, query vs scan.
# Python (boto3): idempotent write with a condition, plus retry with backoff
import boto3, botocore
ddb = boto3.client("dynamodb", config=botocore.config.Config(retries={"max_attempts": 5, "mode": "adaptive"}))
ddb.put_item(
TableName="orders",
Item={"pk": {"S": "ORDER#1001"}, "status": {"S": "NEW"}},
ConditionExpression="attribute_not_exists(pk)", # don't overwrite a duplicate event
)Domain 2: Security (26%)
- Auth: identity federation; fine-grained permissions, tokens and MFA; programmatic access; presigned URLs; signed calls; IAM policies for principals and resources; assuming roles.
- Encryption and sensitive data: at rest and in transit; ACM and AWS Private CA; client-side vs server-side encryption; KMS encrypt/decrypt; masking PII/PHI; Secrets Manager.
- AI service security (new): see the list above.
Domain 3: Testing and Deployment (22%)
- Artifacts: dependencies, container images, AppConfig, Parameter Store and Secrets Manager for environment config, API Gateway stages.
- Testing: tests with SAM, CDK and CloudFormation; mock APIs in API Gateway; JSON test events; Lambda aliases and environments; AI-generated tests.
- CI/CD: Lambda packaging (.zip, container image, layers); ECS, EKS and Fargate deployments; SAM and CloudFormation templates; CodeDeploy; CodePipeline; blue/green, canary and rolling; rollback; branches and labels.
Domain 4: Troubleshooting and Optimization (22%)
- Root cause: debugging, reading metrics, logs and traces, CloudWatch Logs Insights queries, deployment failure logs, AI troubleshooting suggestions.
- Observability: structured logging, custom metrics (embedded metric format), alerts, X-Ray tracing, health checks and probes.
- Optimisation: provisioned concurrency, connection pooling, memory and compute sizing, SQS and SNS fan-out, ElastiCache, DAX and CloudFront caching, finding bottlenecks.
6-week study plan
| Week | Focus | Hands-on |
|---|---|---|
| 1 | IAM roles, SDK, CLI, S3, presigned URLs | Upload/download app with presigned URLs |
| 2 | Lambda, API Gateway, DynamoDB | Serverless REST API with SAM |
| 3 | SQS, SNS, EventBridge, Step Functions basics, retries and DLQs | Order-processing flow with a DLQ |
| 4 | KMS, Secrets Manager, Cognito, AI service security | Call Amazon Bedrock from Lambda through a VPC endpoint with guardrails |
| 5 | ECR, ECS/Fargate, CodePipeline, CodeDeploy, canary and rollback | Container pipeline with automatic rollback |
| 6 | CloudWatch, X-Ray, caching, tuning; AI coding tools; official practice | Fix a slow function using traces; book the exam |
15 scenario questions with answers
These are original scenarios written for this guide, not exam questions.
1. An order API must return quickly while a slow PDF invoice is generated. What pattern?
Asynchronous: API Gateway → SQS (or EventBridge) → Lambda worker.
2. A Lambda consumer sometimes processes the same message twice. How do you make it safe?
Make the handler idempotent, for example with a DynamoDB conditional write keyed on the message ID.
3. Failed asynchronous Lambda invocations must be kept for later analysis. What do you configure?
An on-failure destination or a dead-letter queue.
4. Lambda must read from an RDS database in a private subnet. What is needed?
Attach the function to the VPC subnets and security groups. Use RDS Proxy to pool connections.
5. A mobile app must upload files directly to S3 for 10 minutes without AWS credentials. How?
Generate a presigned URL with a 10-minute expiry.
6. A DynamoDB table is throttled because most writes hit one partition. What do you change?
Use a high-cardinality partition key (or add a suffix to spread writes).
7. You must find all orders for a customer by date, but the table key is order ID. What do you add?
A global secondary index on customer ID and date, then Query (not Scan).
8. Database passwords must rotate automatically. Which service?
AWS Secrets Manager with rotation.
9. Your app sends customer chat text to a foundation model. How do you stop PII and prompt injection reaching the model?
Apply input and output filtering: sensitive-information detection and redaction, denied topics and prompt-attack protection (for example, Amazon Bedrock Guardrails).
10. An AI agent can call a "refund" tool. How do you keep that safe?
Authorise tool use per action and require human-in-the-loop approval for sensitive actions such as refunds.
11. A new Lambda version should get 10% of traffic and roll back if errors rise. How?
CodeDeploy canary deployment on a Lambda alias, with a CloudWatch alarm for automatic rollback.
12. Feature flags must change without redeploying. Which service?
AWS AppConfig.
13. A request is slow somewhere across API Gateway, Lambda and DynamoDB. What finds the slow step?
AWS X-Ray tracing.
14. A latency-sensitive Lambda has cold starts at peak. What helps?
Provisioned concurrency (and right-sizing memory, which also adds CPU).
15. Repeated reads of product data overload DynamoDB. What do you add?
A cache: DAX for DynamoDB, or ElastiCache.
After Developer – Associate
- Practise with AWS serverless interview questions and API Gateway interview questions.
- Go further with the Bedrock engineer roadmap and DevOps Engineer – Professional.
- See live cloud developer jobs.
Cloud Soft Solutions in Ameerpet, Hyderabad teaches AWS development, serverless, DevOps and generative AI on AWS with real-time projects. See the APEX program, the FDE PRO program and all courses.
📞 Call +91 96660 19191 · WhatsApp · Flat No. 513, 5th Floor, Nilagiri Block, Aditya Enclave, Ameerpet, Hyderabad
Frequently Asked Questions
When does DVA-C03 replace DVA-C02?
DVA-C03 is used from 1 December 2026. The last day to take DVA-C02 is 30 November 2026. Registration for DVA-C03 opened on 27 October 2026.
How long is the AWS Developer Associate exam?
130 minutes, with 65 questions: 50 scored and 15 unscored.
What is the passing score for DVA-C03?
720 on a scale of 100 to 1,000.
How much does the AWS Developer Associate exam cost?
USD 150.
What is new in DVA-C03?
AI-assisted development tools, integrating managed AWS AI services, a new task on AI service security, container image builds and deployments, and presigned URLs.
Which DVA-C03 domain has the most weight?
Development with AWS Services, at 30%.
