“I’m still committing most days” - Ken Colton, CEO, on Medal's Engineering-Led Culture

I sat down with Ken Colton, Medal's CEO, to talk about the engineering culture he’s building here.

First, a little background

Ken joined when Medal acquired the company he was working at, where he was a tech lead. He worked  across the stack: on front end user-facing features, their Electron Windows desktop app, and the cloud recording backend that recorded videos using replay files. He spent his first two years at Medal alternating between leading a small team and as an IC before making the jump to Director…then to CTO and now CEO. But still writing code all along the way.

‍

‍

‍

‍

High bar, fast advancement

What does career trajectory look like for an ambitious engineer?

Currently, the org structure is flat, and Ken has 25 direct reports. He does all of the performance reviews himself. He wants every engineer to have someone they can learn from, and the reviews are how he gives guidance on where each person can level up. He doesn't track everyone's work on a day to day, but because he’s always in the code, he maintains a general sense of where everyone is contributing.

"The right person can make almost any jump," Ken said. What Medal needs are people who can ship code and make features. He puts almost no weight on someone's background on paper. At Medal, there's very little correlation between a résumé and how far someone advances.

"If you can make our users happy you'll do well here."

Medal is not the place to learn the fundamentals on the job with a mentor walking you along a linear career path. The bar to entry is higher than at a lot of places, and the expectation is that you hit the ground running. But if you're highly skilled and you've felt in your career like you were better than the people around you and it wasn't being recognized, that's not a problem here. You advance as fast as you can build.

Builders first

Medal prioritizes builders first, particularly people who are product and user oriented. Everyone on the team likes hard technical problems, but the true common motivational thread is that ultimately they're building things people actually use. It's part of why Medal hires people who play games. They don't have to use Medal on day one, but they need to want to look at the product, understand how players use it, and eventually use it themselves. Medal still runs user studies, but Ken wants engineers who already have an intuition for what's a good idea, based on their own experience as players.

Trust is its own hiring priority. Medal wants people who can be trusted to work autonomously and communicate well, and in return they're given real agency over what they work on and how.

The case for a little “chaos”

At a high level, Ken sets the product vision for where Medal is trying to go strategically: like building out game discovery so players try new games and make more clips, or making Medal a place where you build your gaming identity. Each group of designers / engineers takes that and turns it into the actual work.

He also doesn't want any process or structure to be too rigid. Everyone weighs emerging issues and experiments against the known high priorities, and Ken trusts people to sort it out and spend their time where it counts. If a good user idea comes up, or there's a bad bug, it doesn't sit waiting for a slot in the roadmap. If someone says they should really fix something, he wants them to just go fix it. He prefers that to good work stalling over an obsession with where it fits in a roadmap, even if that means things don’t fit neatly into a timeline.

Hence "chaos."

The team structure works the same way. There are both vertical and horizontal groups, and they're not mutually exclusive.

The vertical groups are organized by feature area or outcome. There are groups dedicated to game publisher partnerships, building the developer portal, the promotion system, the game recommendation system, core product features…each of these are cross-functional and consist usually of frontend engineers, backend engineers, and a product designers working together.

The horizontal groups are more informal and more code-oriented: a desktop group, a frontend group, a backend group, each meeting to talk through technical challenges rather than product features. Almost everyone is in at least one horizontal group and one vertical group, and people move between them as projects come and go.

Planning matches that looseness. It's weekly and intentionally not procedural. There’s no scrum master or dedicated sprint plan. Each group decides how it wants to handle its own project management. Some do more planning and scheduling, others just name what needs to get done that week and get it done.

The tech stack

Nobody would look at Medal’s tech stack and be surprised by it. The choices are boring very much on purpose.

Medal is built on Electron and React because they're optimized for speed of iteration and the team knows them well. They're a proven way to ship desktop apps (and in good company with giants like Discord and Slack). The parts that need to be fast are written in C++ and everything sits on standard relational databases.

Ultimately, the team spends time building features instead of chasing new tech because it’s trendy or has hot marketing behind it.

Clarifying General Intuition

General Intuition is Medal's research lab, a separate Public Benefit Corporation. It builds foundation models for spatial intelligence by analyzing Medal’s gameplay clips and the actions and events that happen in them, like "player moved forward" or "goal was scored."

General Intuition just $320M at a $2.3B valuation led by Khosla Ventures with participation from General Catalyst, Eric Schmidt, and Jeff Bezos.

For engineers, the day to day overlap with General Intuition depends on the role. For some it's transparent and they rarely touch it. For others, especially people working on video collection , recording, and in-game action detection, it's a big part of the work.

The incentives line up: whether the goal is more people using the product or more data to train models on, both want to grow, and each pushes the envelope in a way that pulls the other forward. There's also demonstrable mobility between the two, depending on where engineers find their skillsets and interests most applicable. (Our most recent recorder engineer was brought in to do event detection for Medal's auto-clipping and now spends most of his time on General Intuition work.)

Still hands-on

Ken spends a lot of his own time keeping the codebase clean, modern, and simple, so both humans and AI can contribute to it efficiently. As a team scales, and as agents start writing code, code tends to copy the patterns already in the system. The concern is that with how fast code is produced daily, any kind of bad practices that create inefficiencies and poor practice. Bloat can spread like a cancer if not carefully managed, while a clean codebase keeps everyone moving faster. Ken is still committing most days.

AI for better code

Ken wants the whole team using AI dev tools, but with a specific bar: AI should make the code better, not produce the same or slightly worse code faster. AI can lead to more thoroughly tested, cleaner, more well-documented code…often better than the engineer would have produced alone. But ultimately nobody should turn their brain off and accept whatever the model returns. AI doesn't replace the hardest parts of engineering: thinking about how systems fit together and keeping it all clean and replicable. It's there to raise the quality bar, not replace good judgment.

Engineers like it here

Medal is an engineering-led org. There are basically no people whose only job is to tell others what to do, nor any pure project or product managers. Everything is IC and productivity-oriented. There’s no detached business stakeholder reciting KPIs and metrics or demanding roadmap priorities for a product they don't even engage with.

Medal’s leadership is close to the product. They are elbows deep in the platform and play the games. The person setting timelines understands what he's asking for, and the person reviewing your PR will merge it himself.

Ultimately, Medal is the kind of place where the CEO is still committing code at midnight while racking up 600k eSPM in Factorio. That tells you most of what you need to know. 🤷🏻‍♀️

‍