At x.ai we think some things are worth innovating (product!), and some things aren’t (business model). This measured approach extends to how we manage our technologists—in this area we do things a little differently than most startups. We’ve arrived at a management philosophy and a set of tactics that work but that are also constantly evolving as our team grows. This is a snapshot of how we do things today.
A single tech team
We don’t organize by function. Data scientists and engineers in backend, testing and Q/A, frontend, infrastructure, etc all belong to the same team. Self-organized, cross functional teams tackle specific projects. We focus on making impact (pushing code to production), adding value to the company, continuous learning and helping each other grow.
Goals and selecting projects
Having a single tech team solves one problem—silos—but it leaves others: how do you figure out what to work on each quarter, and who does what work? How do you get the right work done most efficiently with the lightest, leanest process?
We do this in two steps. First, we set quarterly company goals. The executive team proposes the company goals in the form of a Request for Comments (RFC) to the whole company. After the comments are discussed and resolved, the goals are settled.
Then a number of self-organized teams work together with the product team to create, scope and prioritize projects within each team that will drive us towards these goals. Team leads alongside with various project leads take on the day-to-day coordination of work. Technologists in these roles leverage their leadership and communication skills to make sure code is delivered; however, they don’t have people management responsibilities and don’t sit in a different hierarchy than their technologist peers.
People managers focus on people and not code
In this context, the key function of a technical people manager is to ensure that our technologists are supported to get the right work done.
Most companies agree that people are their most valuable asset, and yet they don’t dedicate their managers’ full attention to people. Somehow people management is seen to as a part-time gig. We don’t get it.
At x.ai, people managers are absolutely technical but they don’t code. We believe our technologists deserve the dedicated focus of our managers. The converse is also true: we want our expert technologists to make technical decisions, not the managers who are half removed from the task. Merit, not seniority, should drive decision making.
Technologists select their managers
We take self-organization one step further. Our technologists get to select their own managers every six months.
We practice servant leadership. People management is a different role on a separate career track than individual contributors. It’s not de facto a promotion (though we have sourced some of our technical managers from our pool of technologists). While People Managers influence salary adjustments, promotion and separation decisions, they’re never the sole decider. Letting our technologists select their own manager enforces servant leadership and keeps our People Managers accountable to their team. There’s no better feedback than your team’s votes.
Learning and training
Our people managers spend a significant amount of time helping, coaching and nudging our team to perform optimally day in and day out. We strive to deliver feedback that’s brutally honest, kind and in real time. Much of our learning comes through on the job training. Many technologists pair code, though it is not required. People volunteer to give lunch and learn sessions weekly, sharing topics with the whole company (e.g. using Shapeless in Scala, better project management, applying 5-Whys). For deeper topics with high interest like deep learning or data intensive applications, people self-organize into study groups to work through exercises and curriculum weekly.
Finally, the whole company shares a self-managed conference budget. People take money as they need for various conferences, no approval required. The only requirement is to document the withdrawal publicly.
Career development as Tour of Duty
We have no illusion that every employee will be an x.ai lifer. We hope some of them will be. With that, we seek to create “tours” for each individual with defined company impact and growth targets. Scoping each tour to be 2-5 years, we can put more context in career development discussions.
So far the results are pretty amazing (and yes, I’m biased): we’ve built an AI personal assistant who schedule meetings so well, many mistake Amy and her brother Andrew for a human.
If this way of tech people management intrigues you, look us up. We’re hiring!
Like What You’re Reading?
Experience autonomous scheduling
with our AI scheduling assistants