DevEx Is About Making the Car Faster, Not the Driver

About Shahab Malik

Shahab Malik is a UX Researcher at JPMorgan Chase, where he focuses on Developer Experience (DevEx) within the firm’s Internal Developer Platform (IDP). Over the past two years, he’s partnered with engineering, product, and DevOps teams to understand how developers work, what slows them down, and how to build systems that truly support them.

With a PhD in cultural anthropology, Shahab brings both qualitative and quantitative methods to studying human behavior in complex technical environments.

What Does a UX Researcher Do on the DevEx Team?

Most organizations don't have UX researchers in DevEx. They don't even have product managers. Platform engineering teams are generally run by just engineers. UX research in general is the human component of experience, making your machine or your product more user-friendly and engaging for all kinds of users. My role is just focused on engineers. It's trying to make their lives better. I try to advocate not only for better engagement with tooling but also the facilities, the environment, and the culture.

Because when we talk about productivity, that also includes many things around a person's life. You have to eat, you have to sleep, you have to code, you have other engagements, and you have life aspirations. I try to really advocate for what it means to be working in this environment and to create value for the organization and for your customer.

Platform as a Product 

It requires large organizations with budgets to have product managers and UX designers and UX researchers in platform engineering, but that's the way that we feel the best products are built. Treating the platform as a product, and that product requires UX, design, and research, and then engineering.

Metrics Are for Identifying Bottlenecks

The starting point for most organizations is DORA metrics. They’re the bread-and-butter metrics of DevOps environments.We have a lot of metrics - line metrics for system performance throughput, instability, how fast it takes for a review to happen, and those bottlenecks. But the focus should not be a particular metric or a set of metrics but on what the bottlenecks are that your organization is facing.Do you have the right sensors to be able to find those bottlenecks?

You can track multiple metrics like change lead time, throughput, stability metrics, diffs, diffs per day, diffs per hour…it goes on and on and on. It can be all overwhelming. But what you should be doing is trying to find where the bottlenecks are.

DevEx is not Police

The analogy that I make is that DevEx is like the Department of Transportation. 

We have all the cameras and sensors, and we're tracking speed and flow. And we are doing that to know where the traffic is, where the red is on the map, and then try to solve that with resources, make more lanes, or bring in better infrastructure.

We are not the police. They have the same cameras and sensors, but they use that data for a punitive approach. They focus on the individual: that guy's going too fast, that guy's going too slow, that guy's not moving at all. Let's go give him a ticket, or let's go discipline and punish him.

Making the Car Go Faster

Another analogy I like to make is we’re making the F1 car go faster. F1 cars also have all sorts of telemetry and dashboards; some of it is even public, and anyone can see how hard a driver braked on turn one, when they hit the throttle, etc. 

Their assumption is a driver wants to go fast. They treat their drivers like rock stars, and pay them like rock stars, and the question is never how to make the driver go faster. Their focus is, how do we make the car go faster?

We always talk about developer experience, but often the topic of developer productivity creeps up. The answer to making engineers more productive, to making them 10x engineers, is never to track and measure individual performance.

If you're walking, and I have a stick in my hand and tell you that you have to run, I can get you maybe from five miles an hour to 15, but you’ll burn out soon. Wouldn't it be better for me just to give you a moped or a car? What about an airplane? What about a rocket ship?

You're never going to get to Mars trying to whip somebody to run faster. You have to provide a rocket. And the platform can be that rocket. 

Let’s not worry about how fast somebody can run. Let’s assume that they will be fast if they’re in a rocket.