More Resources with Open Source

3 recommendations you should experiment with.

We are constantly searching for tools that simplify and make our job easier. This motivation is much stronger when working in a field as complicated as software development. With so many tasks that must be tackled and executed all at once, finding tools that help automate a couple of the processes.

The good thing is, the open-source community is never lacking tools that work wonderfully in lessening engineers’ burdens. Most digital services nowadays are built on the back of LAMP (Linux, Apache, MySQL, and PHP) and MEAN (MongoDB, Express.js, AngularJS, and Node.js) stacks, which are open-source software. There are also other tools, such as Kubernetes and Docker, which are staples in most organizations’ stacks.

At DKatalis, our engineers actively keep up with new open-source tools available on the pool. Some are also active participants in the community, exchanging tips and hacks. If you enjoy the series of our machine learning engineers’ experiments with MLflow and Dramatiq, here are some other considerations for you!

Feast

Feature stores are a critical piece of Machine Learning infrastructure within the company. One of the core benefits of having a feature store is for Data Scientists and Data Analysts alike to share features, thereby saving time having to recreate features.

Another benefit is that data analysts and scientists become self-empowered, as they can deploy new data assets without an engineer’s involvement, unless on a production level.

Feast, an open-source with an Apache 2.0 license, is one of the available options you should try. With core features that support online and offline feature stores, historical feature retrieval, and feature registry, Feast is also supported by an active and engaging community!

We have covered Feast best practices in this article, containing easy-to-follow setup instructions and a few secret hacks.

Redoc

API documentation is very important for developers, as it hopefully gives clear information they need during their code activity. Because, especially if you’re working in a team, it’s clear that an API will be used not only by its owner but also by other devs in the team. Good documentation will help developers understand how to integrate their app with the platform.

SwaggerHub might be a well-known open-source player for this purpose, but it’s always good to have some options in hand. If you haven’t heard about Redoc, then now you have.

Redoc is an open-source library that generates API documentation with Open API as the single source of truth, which can be written in YAML or JSON and will represent the structure of your API.

Its strength lies in its intuitive, clean, and non-cluttered user interface. Other benefits include easy export to file/document format, seamless deployment, and usage as it supports direct links to GitLab URL, and easy navigation with an embedded search feature. Redoc also creates API documentation in a format that is familiar to developers.

The learning curve is that developers must familiarize themselves with OpenAPI, and that users need to input code snippets manually.

BentoML & Yatai

Imagine having a team of Data Scientists that can self-deploy models.

Well, one way to do it is with BentoML and Yatai. The way the two work is this way: BentoML is a Python open-source library that helps users to create a machine learning-powered prediction service in minutes; then you deploy, operate, and scale Machine Learning services on Kubernetes with Yatai.

The latter helps you configure many things pretty easily, such as the number of replicas and handling CUDA, among others, as shown in this experiment run by our ML engineer. And another plus point is, the BentoML team is pretty active and helpful on their Slack channel, so new users have plenty of resources to learn from.

Have any other recommendations? Hey, you might be a good fit for DKatalis. Join our team of inventive and daring engineers, we’re hiring!