Paul is Systems Architect with a degree in Aerospace Engineering. He started off in aerospace and learned software development. He has many years of experience developing software for organizations like NASA, Symantec, CultureMap, SaltStack, etc and now he’s building AI based planning software for orbital vehicles.
Take your chance to ask him anything now and join for the discussion on September 23rd, 12 pm EST time.
Interesting career path!
What made you change aerospace to software engineering? And what sphere is more open towards innovations adoption? Thanks!
Can we rely on machine algorithms in aerospace field? What dangers automation might cause on space exploration?
Hi! What’s the project you are most proud of?
Thanks for taking the time!
How do you think space exploration will advance in next 30 years?
If you could implement one technology for aerospace without any limitations, what would it be?
What is the difference between software engineering in aerospace field and, say, general tech?
What does the process of managing sattelite operations looks like as of now? What future changes you are working on?
Thanks!
Thanks for doing it!
What are your thoughts on space tourism as a business?
Exciting!
How hard is it to build a career specifically in aerospace software engineering? What set of skills are a must?
When I first started in Aerospace, many engineers didn’t want to do the software tasks. So I started to volunteer for software tasks. I enjoyed it so I decided to do more of it.
Software is more open to innovation. That’s part of why it’s “Soft”. However, aerospace like many other industries is getting eaten by software. So it’s starting to follow the same route.
This is a discussion we have regularly. For our purposes right now, yes, because there is still a human in the loop. Also a lot of my current work is optimization by AI. So a less optimized version isn’t necessarily a total failure.
With that said, as AI advances in the field, AI will have to “show its work” just like any human engineer. So we try to back up our AI predictions with data and use machine learning techniques that let us track the decisions being made. So yes we can rely on AI in space, but like anything in space, it’s complicated.
As for the dangers, it really depends. A bad machine learning algorithm could mean something small like a weather observation gets missed or you receive a download slower. However, if that data is important then it could be pretty big. So hopefully we have resilient and redundant systems.
In the next decade you will see an explosion in orbital vehicles. Microsoft just brought online it’s ground station API this week, and AWS has had one for a couple years. You are also seeing more launches and satellites so these APIs are needed. Space is the next data center battle ground.
Of course corporations will battle for your development dollar, but you may also see countries battle for space supremacy. So you may also see a space cold war evolve.
With all this data and sensors, space will be the next “smart phone”. Just as for Uber to be created it needed the smart phone platform, a new tech unicorn will emerge based on this space data.
A worldwide food delivery service. To get fresh pizza from Italy would be awesome. 30 minutes or less, of course.
It can be much more strict for certain requirements than general tech. Many times you have to document extensively and/or test more formally. And many time there are requirements for certification. So once something is certified, you don’t want to change it or else you go through that process again.
In my space shuttle days, NASA had a requirement for backup software that was independently developed. So there were two “OSes” for the space shuttle, developed by two different companies. The backup flight software was never used in practice. You can develop things that never get used in real life.
With that said, since trust for software is going up, many of the old restrictions are being changed. So many systems can be written today in a faster process. It’s getting closer to general tech but there are still more restrictions.
At it’s most complex end, it is about conflict resolution and managing priorities. But since we are talking about satellites, you have to use orbital mechanics and model a vehicles capabilities so you can determine where conflicts exist. And then that is where machine learning comes in. In a complex environment, we can get the computer to optimize priorities faster than a human can.
In the future, we hope to expand to manage more infrastructure like ground stations, drones, etc.
I wish it would expand. I wanna buy a ticket, but the price is too high! Hopefully we can make some break throughs to make the journey cheaper. Elon Musk and SpaceX are starting to chip away at the problem, but we still have a ways to go.
You could approach it from multiple angles. One, get an aerospace degree and focus on writing some software additionally. This approach is good if you would like to get into writing flight software because that requires knowledge of orbital mechanics, aerodynamics, etc.
Two, get a Computer Science degree and apply for an aerospace job. There is so much software work in aerospace that often generalists are also hired. You probably won’t be put directly on writing flight software but there are so many other related things you can work on. For instance, machine learning, simulation, database, APIs, etc are all used and many of the aerospace engineers do not focus on these.
Three, get any other engineering degree and apply for an aerospace job. Aerospace companies hire mechanical, electrical, etc engineers also to write software. These are all systems that need software on a vehicle. Basically, if you have some technical chops and you can write software, go for it.