top of page
Anaconda Enterprise (Redesign)
Continuum Analytics / Anaconda, Inc
Austin, TX
Anaconda, Inc, formerly known as Continuum Analytics, is a company world-famous for its Python distribution and package management. If you have ever touched Python, it needs no introduction. For anyone else, just know that it is what has enabled beginners and experts alike to use most popular data science programming language in the world.
Anaconda Enterprise is the evolution of what started as a humble distribution and package manager. Before I joined Anaconda in May of 2017, they had already developed several discrete products that leveraged the core principles of Anaconda and its associated tools. Our users, however, wanted a comprehensive suite that encapsulated the entire data science workflow. We also saw our customers shifting into enterprise territory, and they wanted to bring Anaconda and the packages they knew and loved with them, which is an added challenge, given that many enterprise organizations have reservations--some founded, some not--about the use of open-source software in enterprise production environments. And thus the need for Anaconda Enterprise was born.
When I joined the company, Anaconda Enterprise was the skeleton of a data science platform. With weak design input and architectural challenges from cobbling together existing tech stacks, it needed a complete UX overhaul. The navigation and workflow were so confusing that the home page was a dictionary attempting to explain how each part of the platform could be used.
Before:
After:
We scrapped nearly everything and went back to the proverbial and literal drawing board. With four years of experience working with data scientists under my belt--and with internal data scientists, sales engineers, and implementation engineers serving as proxy users--we created a new workflow, UI, and visual language from scratch.
Central to this redesign was the introduction of "Projects." Current Anaconda and conda users typically begin their workflows with an environment, or a collection of packages and libraries that enable a particular experiment or analysis, instead of with an artifact or collection of artifacts themselves. Many of our legacy users might not even consider their work a "project" until the time comes when it needs to be shared, and sharing data science work is no small feat, since it also requires that they share the environment that the experiment utilizes. In large enterprises where collaboration is critical, experiments need to be extensible and reproducible by many different actors. We therefore developed the concept of a project to be the epicenter of data science work. In addition to housing artifacts like notebooks, data sets, and text files (think a simple folder), projects also contain any dependencies (packages, libraries) that the experiment requires, and they can be easily shared with any user in a particular organization.
While data scientists are the primary productive users of Anaconda Enterprise, the platform is also tailored to meet the governance needs of large companies with a host of features for the IT professional, including extensive activity logs; project, environment, and installer versioning; LDAP and AD integration along with customizable permissioning; repository and channel mirroring; and support.
New wireframes for the first release of the new platform were delivered in 1.5 months, and a new visual language was completed two months later.
Disclaimer: Because the old tech stack used a front-end framework that completely rejected CSS, the existing product today looks unfortunately very different to the designs below. After proving the need for a new front-end language to Product Management and Development, we have begun the process of migrating to a new front-end framework so we can implement the full intent of the designs.
First Iteration:
Latest iteration:
bottom of page