Avatar of Jake CooperJake Cooper

How We Work (Volume III)

This is the third post in the “How We Work” blog series. In Volumes 1 and 2, we focused on why you’d love working at Railway. In this one, we’re gonna focus on why you might not.

At Railway, we believe in leverage. Whether it’s product, process, or people, leverage is at the core of everything we do.

We’re ~20 people, powering >650k users, >2600 businesses, 24 hours a day, 7 days per week, 365 days a year.

And, in 4 years, ~20 is about the number of people that we’ve found who operate the way we do.

So, it looks like we’re in search of rare people.

What kind of people exactly? I’m so glad you asked.

For starters, they’re highly technical, highly autonomous, and highly driven. Usually this archetype carries the risk of the person being... well... an asshole.

Because we actively choose not to work with assholes, we also look for people who are high EQ and low ego.

This intersection of the Venn diagram, as it turns out, is really, REALLY hard to find. Especially when most companies, startup or not, trend towards putting people in a box. Because boxes are predictable, and companies need predictability to scale.

In most companies, they manage first, and build leverage second (if at all).

At Railway, we build leverage first, and manage second (if at all).

But this isn’t actually that new — all we’re doing is …

We’re building a high signal, low noise organization (Musk). We’re building an organization where we hire people smarter than ourselves (Carnegie), give them as much freedom as possible (Simons), get out of their way (Lacoppa), tell people what we need, not what to do (Drucker), and, instead, let them tell us what to do (Jobs).

So, if many of the greatest business titans of the last two centuries have the same advice, why don’t more companies do this?

As it turns out:

  • Finding these kinds of people is hard
  • Building an org for them is harder
  • Scaling an organization full of them is harder still

So, after 4 years, I’m gonna cheat a little. I’m gonna tell you exactly what we’re looking for, exactly what Railway is like internally, and you can decide if Railway is the place for you.

In our last two posts, we’ve preached the gospel of remote work and spread the sermon of how we work. Cool projects, lots of freedom, flexible working hours, unlimited PTO, and basically infinite freedom to build what you love.

But there’s a couple things that are very different from most companies that are worth flagging.

In most other companies, planning is top down. Your projects are assigned, and your PM makes sure to cut everything into bite sized chunks so that you don’t choke on them.

This doesn’t really fit with “treating people like adults.” So we do it a bit different.

Yearly, we set the direction for the company, AKA “What has to happen to hit our milestones.”

Quarterly, we set goals for ourselves and the company, AKA “What has to happen now.”

P0 = “Have to do now or the company risks dying.” P1 = “Blocks further quarter’s work.” P2 = “Nice to have.”

P0 = “Have to do now or the company risks dying.” P1 = “Blocks further quarter’s work.” P2 = “Nice to have.”

Once we agree what has to happen, we work towards our shared goals and expect others to in tandem. People have freedom to set projects in domains they’re passionate about, assure those projects meet our goals, and are expected to hit those goals autonomously.

While you might once in a while get some help breaking down tasks for your own planning, there’s no “PM on tap.” You’ll work directly with customers to ideate, design, ship, scale, monitor, and improve on what you believe will move the business forward.

You will be directly responsible for your projects and their outcomes.

You will, in essence, have more than enough rope to do whatever you need. We aspire to take the limiter off every person at this company and see where the open road takes us.

In most other companies, when you want to give someone feedback, that feedback flows from you, to your manager, to their manager to them. Throughout this game of telephone, the feedback often gets watered down, twisted up, or just straight up dropped on the floor. You also have no idea which because it’s a black box.

Worst of all, in some situations, feedback ends up being given far too late. It’s silence, silence, irredeemable PIP. This is not how we operate. We treat people like adults, and we are direct with each other in airing, and resolving, our grievances.

We follow a “kind but direct” principle. You will receive feedback directly from other individuals at the company. Sometimes this will be in private, but it’s encouraged to move towards a more public format, such that others can know you’re working on it and even help you improve.

Some public feedback Ray gave me a couple months ago when I sprung on-call on the design team.

Some public feedback Ray gave me a couple months ago when I sprung on-call on the design team.

Very quickly upon starting, you will probably receive feedback from your peers.

In most companies, it’s because they want to build a case for PIP.

For us, it’s because we know you want to be great.

One of our coworkers said it best: “A guided torpedo arrives at its target by making a series of mistakes and continually correcting its course. Be the torpedo.”

Most companies operate on a “Pull based” model. “How’s this project going?” “You got that TPS report?” Etc.

This results in a variety of failures modes:

  • Much like in “Office Space,” many different people can ask you about the same thing
  • Management, who need this information to do their job, may ask for it at inopportune times
  • Coordinating meetings in a company spanning every timezone straight up doesn’t work

Instead, we push our information. Daily wraps, project channels, and (my favorite) #progress-pics. All in public, so everyone’s aware.

Most recently we’ve been sharing, over text, what our “big rock” is for the week. And at the end, we close it out, or say why we haven’t.

Our “Weekly Big Rocks” thread for the Platform team.

Our “Weekly Big Rocks” thread for the Platform team.

This solves every single one of the above problems AND reduces managerial overhead required to synchronize the system because everything’s in the open and in one place.

As a result, I spend very little time “managing” and a lot of my time coaching, collaborating, and helping people ring the bell of greatness.

Most companies hire for redundancy. You’ll work on whatever is pertinent to the business, regardless of your skills or interests. The risk of working on something “out there” is heavily outweighed by the organization’s inability to replace the cog should anything go even the tiniest bit off-track.

At Railway, we hire for obsolescence. We seek to hire people who will kick our ass at something, figure out what they’re about, and find the fastest possible way to have them own it.

As an example, Miguel joined us almost a year ago. Within his first few days, he zeroed in on the networking stack — specifically, this ticket about DDoS mitigation:

One of the first messages Miguel sent in the company Discord bringing up PLAT-432 (”RFC for DDOS Mitigation”).

One of the first messages Miguel sent in the company Discord bringing up PLAT-432 (”RFC for DDOS Mitigation”).

Within weeks he’d shipped the v0 of our WAF rate limiting system called “Abuse Daemon.”

Within months he’d shipped Scale-to-Zero (aka “serverless anything”) during our launch week.

Fast forward to today, almost exactly a year later, he’s now the primary owner of our entire networking stack, including the custom proxy we released last week to replace Envoy.

For Miguel and many others at the company, it goes beyond ownership:

"I’m not building our edge network; I’m making systems that aim to take on the status quo for edge networking tools. My motivation isn’t deadlines, it’s what can we do better than anyone else in the world. What bit of magic can we add that will separate our product in unique ways. I feel like I have the trust and freedom to go and try and do the best work of my career”


Here’s the real kicker. By his second month, we had mostly stopped telling him what to do.

That’s how onboarding works here:

  • We hire you because we believe you have the desire and capacity be better than any of us in a vertical you’re passionate about
  • We do 6 weeks of onboarding:
    • 2 weeks of “Here’s 5 straightforward Linear Tasks”
    • 2 weeks of “Here’s 2 problems we need solved”
    • 2 weeks of “What do you think the company needs?”

We then spend our time focusing on how to make you untouchable in that domain. Meanwhile, we make sure to rotate people through zones, and assure the presence of strong run-books and rotational knowledge transfer to mitigate bus factor.

In this sense, every single person you work with is simultaneously better than you, and you can teach them something. You teach something, you learn tons. Perpetual learning machine.

New people excel in depth/slope, and more tenured people excel in breadth simply due to their context from having built prior versions of the systems. This cycle is then repeated infinitely.

For this, a few key things must happen:
  • We hire people who we believe can be better than ourselves in a domain
  • We have the desire and capacity to push ourselves to be great
  • We take risks to push the boundaries of the area in which we seek excellence
  • We, upon completion of our “v0,” part-out our zone by repeating step one
  • We repeat the above until we feel really dumb. For us, there is no higher honor.

    In most startups, you work 24/7. No work life balance. Burnout City. Always 100% on.

    In most big companies, you have tons of perks to keep you there. Pingpong tables, Happy Hours, “Slack as an infinite dopamine mill.” Plus, the org gets in your way, so it’s difficult to be 100% on.

    At Railway, we believe there’s a third option — we work hard, and then we go home.

    Revolutionary, I know.

    We treat work like work, and time off like time off. We choose to not occupy the “middle 50% pingpong table space.”

    If you’re working, be on.

    If you’re not, log the fuck off.

    Monday to Friday, you will see people online at various hours. Maybe the person prioritized spending their morning in the park with their friends, their afternoon spending time with their kid after school, or, maybe it’s 2AM, they’re a night owl, and they’re wrist-deep in the man pages trying to eek out that last bit of magic.

    Saturday and Sunday, barring pages/oncall, it’s silence, and it’s golden. I can count the number of times I’ve personally asked anybody to work weekends on one hand. It’s zero.

    We push ourselves, and then we rest. Because you need both to be great, and we choose to work with great people because, as they push themselves to be better, we in-turn are pushed to be better ourselves.

    There is nothing quite like watching someone break their skill ceiling in real-time.

    It should be noted that, rarely, some people do work weekends.

    Beyond discouraging it, there’s only so much we can do.

    We do treat people like adults after all 🤷🏻‍♂️.

    All of the above, plus many other things, means that, hopefully, Railway is a Great™️ place to work. Great™️ because the people are passionate, the problems are difficult, and because we truly revere the people that we work with and find ourselves inspired by what they do. When you see #progress-pics light up like a Christmas tree, you start to believe the impossible is possible. That’s why it’s great.

    We know it’s not for everyone, but for us we wouldn’t have it any other way.

    That’s how we work.

    If you’re like us, checkout our careers page. We’d love to have you at our little company.