Scaling music production experience for Epidemic Sound.


— Epidemic Sound scaled up from being a small—medium sized music company to be a technology driven company in a fast span. Music intake before this transformation was split between different tools like Slack, Google drive and airtable. There was no single source of data for the company resulting in intense amount of manual labour, errors and general lack of hygeine. All of this made it extremely hard for the company to plan, review and produce more music in less time.


Early workshop involving meeting different departments in the company to define the entire track lifecycle from planning internally to releasing to the world. It also gave us a rough estimate of the problem we were looking at, and what we re solving with Kosmos.


The hypothesis I worked on was simple, if we have a single source of data for our music and our musicians, the company can support production a larger volume of music without deteriorating the quality of our music. This became the basis of our new product.


A very simple design system to help focus on the job.

— We initially built our music production tool with Material UI to focus on the technical challanges in hand. It didn't work since we spent more time fixing the Material UI related layout issues than building the product. Hence, we built a very simple design system (Epidemic Sound didn't have a design system at that time) with a strong focus on less but highly repetitive elements to ease out difficulty for the developers to build the product and focus more delivering to users fast.


Early releases we made were using material ui, due to lack of a design system, which allowed us to push out the early versions of the application fast. It came with a side-effect that, we spent a lot of time fixing the complex material ui components than actually shipping the product.


As a side-project over the summer, I and two more engineers from our team, who were fed up fixing material-ui related bugs built a very-very simple and minimal design system and we phased out material ui in a couple of months after that.

Focus on testing different design patterns.

— We focused very heavily on repeating similar patterns for the users to get familier with the product extremely fast. An example of such a pattern is below.


We divided the experience into different cards based on the types of users interacting with it. For example, users responsible for musician ino and meta-data are different. Then, we tested 2 layouts with our users, a single save button as the only save button for the entire experience and different save buttons for each sections. First section was the favorite amongt the users.

This is an ongoing project.

Since working with music intake involves me dealing with sensitive data, I can't share a lot of information. I will although share a lot more as the project progresses.