Amazon CodeGuru and CodeWhisperer for Java Developers on AWS
summary
Demonstrates Amazon CodeGuru Reviewer (and Profiler) plus CodeWhisperer applied to Java serverless code: CodeGuru flags hardcoded credentials, S3 clients created per Lambda invocation, deprecated SDK calls, deadlocks and MD5 usage and surfaces 22 recommendations from an example repo; Profiler points to CPU hotspots and excessive logging; Lambda scanning via Inspector runs post-deploy. CodeWhisperer generates useful class stubs and algorithms but produces wrong unit-test suggestions and misinterprets requirements (e.g. palindrome/ digit range), so results are mixed and full-repo scans cost about $10 per scan requiring supervision.
AI-written summary · about
chapters (19)
- 0:00Introduction and background
- 1:08AWS CodeGuru overview
- 3:15CodeGuru findings examples
- 7:49CodeGuru concurrency checks
- 9:17Incremental CodeGuru reviews
- 10:01AWS Lambda code scanning
- 12:52CodeGuru integration options
- 13:12CodeGuru Profiler setup
- 14:59AWS SDK version findings
- 15:39CodeGuru cost considerations
- 16:01AWS CodeWhisperer intro
- 17:27Simple functions with CodeWhisperer
- 18:53General tasks in CodeWhisperer
- 19:57Thread-safe Singleton example
- 20:40Complex algorithms with CodeWhisperer
- 23:50Web development with Spring Boot
- 29:06Serverless app generation
- 30:11CodeGuru language support
- 30:53CodeWhisperer vs GitHub Copilot
related talks
35mBring the power of machine learning to your developer operations
35mBring the power of machine learning to your developer operations
02mAWS Community Spotlight: Amazon CodeWhisperer
49mGithub Copilot vs. Amazon CodeWhisperer For Java Developers
14mAmazon Q Builder: Securing your code
19mFrom code to insight: Amazon Inspector & AWS Lambda in action
22mIncreasing developer productivity with Generative AI
48mFrom IDE to production: Actionable ML-powered code security analysis