We're moving from Java to Kotlin with our Microservice Stack (Spring Boot) because it is excellently supported by framework and tools and the learning curve is not very steep Kotlin is way more straightforward and convenient to use while providing less boilerplate and more strictness, which finally leads to better code, which is more readable, maintainable and less error-prone. We especially like Kotlin's (functional) data structures, which are, e.g. compared to Scala, easier to understand and don't require deep knowledge in functional programming.

I am working in the domain of big data and machine learning. I am helping companies with bringing their machine learning models to the production. In many projects there is a tendency to port Python, PySpark code to Scala and Scala Spark.

This yields to longer time to market and a lot of mistakes due to necessity to understand and re-write the code. Also many libraries/apis that data scientists/machine learning practitioners use are not available in jvm ecosystem.

Simply, refactoring (if necessary) and organising the code of the data scientists by following best practices of software development is less error prone and faster comparing to re-write in Scala.

Pipeline orchestration tools such as Luigi/Airflow is python native and fits well to this picture.

I have heard some arguments against Python such as, it is slow, or it is hard to maintain due to its dynamically typed language. However cost/benefit of time consumed porting python code to java/scala alone would be enough as a counter-argument. ML pipelines rarerly contains a lot of code (if that is not the case, such as complex domain and significant amount of code, then scala would be a better fit).

In terms of performance, I did not see any issues with Python. It is not the fastest runtime around but ML applications are rarely time-critical (majority of them is batch based).

I still prefer Scala for developing APIs and for applications where the domain contains complex logic.

We have a lot of experience in JavaScript, writing our services in NodeJS allows developers to transition to the back end without any friction, without having to learn a new language. There is also the option to write services in TypeScript, which adds an expressive type layer. The semi-shared ecosystem between front and back end is nice as well, though specifically NodeJS libraries sometimes suffer in quality, compared to other major languages.

As for why we didn't pick the other languages, most of it comes down to "personal preference" and historically grown code bases, but let's do some post-hoc deduction:

Go is a practical choice, reasonably easy to learn, but until we find performance issues with our NodeJS stack, there is simply no reason to switch. The benefits of using NodeJS so far outweigh those of picking Go. This might change in the future.

PHP is a language we're still using in big parts of our system, and are still sometimes writing new code in. Modern PHP has fixed some of its issues, and probably has the fastest development cycle time, but it suffers around modelling complex asynchronous tasks, and (on a personal note) lack of support for writing in a functional style.

We don't use Python, Elixir or Ruby, mostly because of personal preference and for historic reasons.

Rust, though I personally love and use it in my projects, would require us to specifically hire for that, as the learning curve is quite steep. Its web ecosystem is OK by now (see https://www.arewewebyet.org/), but in my opinion, it is still no where near that of the other web languages. In other words, we are not willing to pay the price for playing this innovation card.

Haskell, as with Rust, I personally adore, but is simply too esoteric for us. There are problem domains where it shines, ours is not one of them.

We needed to incorporate Big Data Framework for data stream analysis, specifically Apache Spark / Apache Storm. The three options of languages were most suitable for the job - Python, Java, Scala.

The winner was Python for the top of the class, high-performance data analysis libraries (NumPy, Pandas) written in C, quick learning curve, quick prototyping allowance, and a great connection with other future tools for machine learning as Tensorflow.

The whole code was shorter & more readable which made it easier to develop and maintain.