H&S Conclave #48. AI solution for mass test generation
Writing a test using AI is not a difficult task today. It is quite another matter to scale this process to a product consisting of dozens of components in different programming languages, with more than 10,000 test cases.
Let's look at the report:
• how the solution works: the Python and Anthropic Agent SDK system analyzes the source code itself, collects a database of use cases, sets priorities, checks coverage, and generates missing tests.;
• at what scale it works: 28 components in Python, Go, PHP, JavaScript and Rust, more than 11,000 test cases;
• what the operation showed: the results of several months of work in real conditions, design and implementation details, difficulties encountered along the way.
Speaker: Dmitry Vinokurov, Staff ROCKET Engineer. In development since 2009, specializing in Linux, Python, QA and AI.
Format: report and discussion, about 1.5 hours, Google Meet. The meeting is being recorded.
Hard&Soft Skills Conclave is a series of open reports on applied technical topics for developers, technical specialists and architects.