The event—the culmination of a dangerous five-year journey since the Juno spacecraft launched into outer space in 2011—was not only a victory for NASA, which broadcast their celebration on Monday night, but for a certain Fortune 500 company: defense and aerospace contractor Lockheed Martin (LMT), which was reveling in the moment behind the scenes. It was Lockheed Martin, after all, that had spent the past 13 years on the Juno mission, working to get the spacecraft to Jupiter on time and on an amazingly tight government budget: It built Juno, tested it to withstand space conditions so harsh they defy the imagination and wrote the computer code that controls the spacecraft. Lockheed knew the stakes, declaring on its website about the mission, “If this doesn’t happen just right, there will be no second chance.”

And on the evening of July 4, Lockheed Martin finally received confirmation that Juno has safely entered into Jupiter’s orbit—where the company will continue to oversee the spacecraft until its mission concludes in early 2018. “Tonight, 540 million miles away, Juno performed a precisely choreographed dance at blazing speeds with the largest, most intense planet in our solar system,” Guy Beutelschies, Lockheed Martin’s director of interplanetary missions, said in a statement.

“It was a pretty exciting time,” Kevin Rudolph, Lockheed’s lead systems engineer for the Juno spacecraft, tells Fortune of the big night.

Engine cut off! #Juno is in orbit around #Jupiter! pic.twitter.com/l4zCLIOC1x

— Lockheed Martin (@LockheedMartin) July 5, 2016

It was the most recent phase of Juno’s mission that was the most harrowing, as the unmanned spacecraft had to penetrate extreme radiation zones surrounding Jupiter. To do so the probe had to be moving at a speed of 130,000 miles per hour in order to get close enough to orbit the massive planet, which is the biggest in our solar system.

On top of the physical risks, Lockheed and NASA were largely flying blind during the final, and most dangerous, four days of the journey, having decided to turn off all nonessential functions of the spacecraft’s electronics, including its photo and video cameras, so that its computer would not be “too busy” or “stressed” by memory or processor usage. It’s the same way a laptop might run slower or crash when too many applications are open, Rudolph says. (See below for NASA’s Juno probe video made before they turned off the cameras.)

In fact, since last Thursday through July 4 at almost midnight, Juno had essentially been on autopilot, running on a single line of code that Lockheed wrote to guide it through its maneuvers, as the spacecraft was too far away at that point to communicate with it in real-time from Earth. Radio signals would have taken 48 minutes each way.

So around 8 p.m. on Monday, Rudolph and his team knew that Juno was getting into the riskiest part of its Jupiter approach, but all they could do was hold their breath and hope it would all be OK. “At that point that’s when the worry level really started picking up,” Rudolph says. “That was when the rubber met the road and we started getting into that radiation environment and we were all pretty tense.”

Lockheed couldn’t predict with absolute certainty exactly how high Jupiter’s radiation levels would be, but it used NASA’s estimations to calculate the likely dose the spacecraft would face—then doubled their calculations just to be safe. To protect Juno’s computers against the radioactive elements, Lockheed built a one-meter cubic vault made out of titanium half an inch thick and put all the spacecraft’s electronics inside it, where the radiation dose would be more than 1000 times smaller than it would be outside the box.

Still, Lockheed wasn’t able to perfectly simulate the conditions Juno would face in its on-the-ground testing. “We know that radiation can affect computers—it can make them work badly or it can make them fail,” Rudolph says. “We had designed this computer to be shielded, but you never really know until you get there and see it work.”

Besides the atmospheric challenges, the mission had “severe cost constraints,” Rudolph says. The budget for the project was a relatively lean $1.1 billion. By comparison, NASA’s first and only other mission to Jupiter, the Galileo, had cost about $1.5 billion—and that was in the 1990s.

Finally, Lockheed and NASA received the confirmation it was waiting for, in the form of a tone (a specific musical note), signaling that Juno had completed its 35-minute engine burn, slowing it down enough to get captured by Jupiter’s gravity and put it into orbit, just as Rudolph’s team had planned (even the length of the engine firing was within one second of Lockheed’s forecast—off by much less than 1%, which still would have been acceptable, Rudolph says). “Everything worked way better than expected,” he says. “That’s when we all breathed a sigh of relief and looked to the ceiling and wiped the sweat off our brow and then we were happy.”

NASA, too, was pleased with the way Lockheed had executed. “At that moment, all that went through my mind was, ‘Wow this thing was perfect. These engineers are amazing!'” Scott Bolton, the Juno mission’s principal investigator, said at a press conference afterwards.

Next, Lockheed will need to send new instructions to Juno for another engine burn maneuver that will slow it down from its current orbit of 53.5 days around Jupiter to just 14 days, where NASA scientists will be able to collect the data they need on Jupiter. Why didn’t Lockheed just slow Juno down to its final orbit speed the first time around? It had never done a test run to see if the spacecraft’s engine could fire continuously for as long as would be required to bring it down to speed. “We couldn’t be absolutely, utterly positive that the engine would work that long,” Rudolph says.

On the other hand, they were “positive” that it could fire in two shifts of 35 minutes or less. “We never want to do anything that hasn’t been proven before,” he says.