The Whiskey and the Light
I was sitting on a rock in the Shenandoah Valley, sipping Catoctin Creek rye, when Dan Roelker said it: "If you can control light, you can control space."
He wasn’t quoting NASA. He wasn’t reading from a slide deck. He was just... thinking out loud. The whiskey was warm, the air cool, and the stars already starting to blink through the trees. He didn’t look like a man who’d led software for Starlink or hacked DARPA’s most classified networks. He looked like someone who’d just finished a long hike and realized the whole damn universe is made of photons.
That’s the thing about Roelker. He doesn’t talk about satellites. He talks about light. Not as a tool. Not as a signal. But as the thing that binds everything together—the quiet, invisible thread between a telescope in Michigan and a probe orbiting Psyche, 290 million miles away.
He didn’t say it with grandeur. Just a quiet nod, like he’d been waiting for someone to ask.
And now, here we are.
The Hacker Who Saw the Code in the Sky
Roelker’s first hack wasn’t a website. It was a router.
At 17, he cracked the firewall on the Johns Hopkins Applied Physics Lab’s internal network—not to steal data, but because he noticed the telemetry streams from a satellite were leaking into the lab’s internal weather station logs. He wrote a script to reroute them. He didn’t tell anyone. He just... fixed it.
That’s the pattern. He doesn’t break things. He notices what’s broken and quietly fixes it. By 22, he was at Sourcefire, building the first real-time intrusion detection system that didn’t rely on signatures. Cisco bought them for $2.7 billion in 2011. Roelker walked away with a check and a quiet sense of disappointment.
"They turned it into a product," he told me. "I wanted it to be a conversation."
That same year, he became the youngest program manager in DARPA history. He ran Plan X—the first publicly acknowledged U.S. offensive cyberwarfare initiative. He didn’t want to blow up networks. He wanted to see them. To map every packet, every flicker of traffic, like constellations in a digital sky.
"We were trying to make cyber warfare visible," he said. "But the truth is, we were just trying to see the light."
He didn’t leave DARPA because he got bored. He left because he realized the battlefield wasn’t in the server room anymore. It was in orbit.
League of Legends, Then the Rocket
After DARPA, he went to Riot Games.
No one saw that coming.
He didn’t join to make money. He joined because League of Legends had the most complex real-time distributed system he’d ever seen—20 million concurrent players, each with their own latency, their own ping, their own little universe of actions happening in real time. He called it "a planetary-scale network of human decisions."
He led the team that redesigned the matchmaking algorithm to handle regional latency spikes without crashing. He didn’t care about the loot boxes. He cared about the flow.
"You’re not playing a game," he told his engineers. "You’re playing a symphony of delays."
He left in 2015, not for a bigger salary, but because SpaceX called. They needed someone to fix the Falcon 9’s flight software after the explosion in Cape Canaveral.
No one else knew how to make software that could survive failure. Roelker did.
He didn’t just recover the code. He rewrote the entire architecture to be self-healing. He called it "the rocket’s nervous system." And when the next Falcon 9 landed, it didn’t just touch down—it remembered how to land.
He went on to lead Crew Dragon’s software, then Starlink’s constellation control system. He didn’t just manage code. He managed the choreography of 5,000 satellites moving in perfect, silent formation across the night sky.
"We didn’t launch a network," he said. "We launched a constellation of eyes."
And then, in 2022, he walked away again.
The Crypto Detour That Wasn’t
He didn’t join OpenSea because he believed in NFTs.
He joined because the platform was built on a decentralized ledger that couldn’t scale. The team was trying to store image metadata on-chain. He looked at it and said, "You’re storing light as a hash."
He spent six months rebuilding their backend to use off-chain storage with cryptographic verification. He didn’t care about the art. He cared about the proof. The integrity. The way light, once captured, must never be altered.
"If you can’t trust the light," he said, "you can’t trust anything."
He left just before the market peaked. Not because he thought it was a bubble. But because he realized he’d already seen the future.
He’d spent years building systems that could see, track, and interpret light across vast distances. He didn’t need to sell digital art. He wanted to collect the light of the stars.
Observable Space: Where the Ground Becomes the Telescope
In October 2022, he co-founded OurSky with Connor Poole, ex-Bird scooter engineer. Their idea? Build a ground-based optical network that could see satellites the way a radar sees planes.
They didn’t want to build more satellites. They wanted to see the ones already there.
A year later, they merged with PlaneWave Instruments, a Michigan-based company that made high-precision telescopes for amateur astronomers and university labs. The merger wasn’t about scale. It was about vertical integration.
"We needed to control the entire chain," Roelker said. "From the lens to the light to the algorithm that sees it."
They didn’t call it a startup. They called it Observable Space.
Today, they’re building 200mm multi-spectral imagers called Iguana—self-contained, off-the-shelf optical payloads that can be launched on any spacecraft and immediately start mapping space debris, tracking satellites, or observing deep-space phenomena.
They’re also building ground stations that don’t need power grids. Solar-powered, AI-driven, deployable in under 48 hours. The U.S. Space Force calls them ATOMS—Deployable Attritable Optical Systems.
"Attritable?" I asked.
"Yes," he said. "If it gets hit by space junk, we don’t cry. We just build another."
They’ve got 175 employees now. Manufacturing in Detroit and Los Angeles. And a $94 million IDIQ contract from the Space Force—the largest sole-sourced contract in the agency’s history.
They’re not building the future of space. They’re building the future of seeing.
The Lasercom Revolution
The most astonishing thing about Observable Space isn’t the telescopes.
It’s the lasers.
In October 2024, NASA used the Hale Telescope at Palomar to communicate with the Psyche spacecraft—290 million miles away—using lasercom. The data rate? 200 Mbps. That’s 100 times faster than radio.
But here’s the kicker: the signal didn’t come from space. It came from the ground.
Observable Space built the receiver.
They didn’t just build a telescope. They built a receiver that could lock onto a laser beam moving at 40,000 miles per hour, 290 million miles away, and hold it steady long enough to read the data.
Then came Artemis II.
The Orion spacecraft, hurtling around the moon, sent a laser signal back to Earth. It was picked up by a ground station in Australia, built by Observable Space, in partnership with the Australian National University.
Data rate: 260 Mbps.
"That’s faster than your home Wi-Fi," Roelker said, grinning. "And it’s coming from a spacecraft that’s not even in orbit yet."
The implications are staggering. No more bandwidth crunch. No more waiting hours for a single image from Mars. In the next decade, every satellite, every probe, every lunar lander will communicate via laser.
And Observable Space is building the infrastructure to make it happen.
"Fiber optics built the internet," says Shahin Farshchi of Lux Capital. "Free-space optics will be the backbone of orbital infrastructure."
Roelker doesn’t need to say it. He’s already done it.
The Argus Array: 1,200 Eyes on the Sky
Eric Schmidt’s Schmidt Sciences is funding something called the Argus Array.
It’s not a single telescope.
It’s 1,200 small, automated, solar-powered optical sensors, scattered across the globe—from the Atacama Desert to the Antarctic ice sheet.
Each one is an Iguana imager, mounted on a mobile platform, capable of tracking 1,000+ objects per minute.
They’re not for astronomy. They’re for space domain awareness.
Right now, there are roughly 3,000 active satellites in orbit. By 2030, there will be 15,000.
And half of them? They’re not talking to anyone.
They’re dead. Or broken. Or spinning out of control.
The U.S. Space Force can’t track them all with radar. Radar doesn’t see small objects. It doesn’t see debris the size of a coffee cup.
But light? Light sees everything.
Observable Space is building the Argus Array to see it all.
"We’re not trying to prevent collisions," Roelker told me. "We’re trying to make the sky visible again."
He’s not talking about satellites anymore.
He’s talking about the sky.
And he’s not alone.
The Department of Defense’s Assistant Secretary for Space Policy, Jeremy Verbout, put it plainly: "The Department is acting on the urgent need for mobile, off-grid robotic telescopes."
Roelker didn’t invent space domain awareness.
But he made it real.
The New Space Race Is on the Ground
The old space race was about who could get there first.
The new one? It’s about who can see the best.
Roelker doesn’t care about Mars colonies or lunar bases.
He cares about what’s already up there.
And he’s right.
We’re not running out of space.
We’re running out of sight.
Every day, more satellites go up. More debris. More junk. More signals. More light.
And most of it? We can’t see it.
Observable Space isn’t building rockets.
They’re building eyes.
And they’re building them on the ground.
"The new space race," he says, "is going to be on the ground."
He means it.
Not in the sense of competition.
But in the sense of responsibility.
We’ve launched thousands of machines into orbit.
Now we have to watch them.
And if you can’t control the light?
You can’t control the space.
He takes another sip of whiskey.
"I don’t know of a bigger question we can answer as humans," he says.
"How do we see what we’ve made?"
And then, quietly:
"And what do we do with what we see?"
The stars are brighter now.
The valley is quiet.
And somewhere, 290 million miles away, a laser is blinking back.