Developers will be developers!

From my past 20 years into IT and software development, one thing I realized that it is the fundamental rights of most of the developers is to keep the bugs in the system (of course without any intention). As we grow as a developers, we start learning design patterns, clean code practices etc., but all things are generally kept aside due to work pressure, quick delivery and many more reasons. And every (almost) developer has common excuses to defend their design/code.

Though we have code quality tools to identify and verify the quality of code at design or run time, why bugs still exist knowing the fact that underlying framework or runtime (CLR/JVM etc.) remain the same. As most of these tools are based on some pre-defined rules and developers are smart enough to manipulate/suppressing these rules in these tools because tools are just the tools. In almost every release, development teams are spending days and nights to fix the last minutes issues to meet the delivery and the question is why this pattern is so consistent and can't we do something to overcome knowing the fact that existing tools and processes are unable to make the software bug free.

There is enough buzz on AI/ML in almost every field, but I don't see much tractions on how we can use the power of AI/ML to make the software bug free because "Developers will be Developers" and as I said they have their fundamental rights to keep bugs in the system. I performed a quick search on this topic and found some interesting articles and thought to develop something that can be injected as like other libraries into software development cycle that predicts the potential bugs (as the engineering fundamentals remain the same for any type of application) while writing the code and suggest the code snippets based on best practices and data collected from past experience.

Most of the current IDEs like Visual Studio, Eclipse etc. are coming up with their own intelligent way to detect code but there is a need to think beyond and build a cross platform library to understand the underlying framework/platform and predict the bugs at every stage of SDLC. Per my understanding most of the big giants like Microsoft, IBM and Google are using some home grown stuff but could not get much detail.

Please do share your thoughts and to start with, I have started an open source project [https://github.com/ML4Dev/BugPredictor] to collect ideas and build something for our development community. You are most welcome to join this community driven initiative.