Django in Space with Loes Crama
Published October 21, 2026
This video features Loes Crama at DjangoCon Europe 2026 in Athens, Greece.
TALK: How Django is helping to build the biggest X-ray observatory to date by Loes Crama
The European Space Agency (ESA) is leading the development of NewAthena (the New Advanced Telescope for High ENergy Astrophysics), an X-ray observatory designed to explore cosmic phenomena such as black holes, neutron stars, and galaxy clusters. The telescope utilizes a 12-meter focal length mirror assembly composed of 600 mirror modules. This talk will explore how Django serves as the software backbone for managing this data. It presents how tracking systems for scientists and engineers are built using Django’s standard core tooling, as well as additional packages where needed.
More information: https://2026.djangocon.eu/speakers/CVHGWK/
Follow DjangoCon Europe on social media:
*X - https://x.com/DjangoConEurope/
*LinkedIn - https://www.linkedin.com/company/djangocon-europe/
*Mastodon - https://fosstodon.org/@djangoconeurope
#django #djangocon #djangoconeurope #djangoconeurope2026
Automatically transcribed, so expect mistakes in names and technical terms.
Speaker 1: Welcome back. It's great to have you here. I hope you all cheered loud and clearly on the photograph. And you energized now because uh it's my pleasure to introduce Luz Klama and um she's gonna be talking about arrays and what you can do with arrays. I I don't know if you ever used arrays, but you can do immensely precise things with them and I'm I'm sure we'll all be very surprised what you can do. She works with the projects with ESA, the European Space Agency. And yeah, I believe You like space? Aries takes space.
Speaker 2: I do, yes.
Speaker 1: Space is huge and I'm really I'm really looking forward to see what we will tell us about space Phase. Give her a warm welcome.
Speaker 2: Hi, so my name is Luz and today I will be talking about how Django is helping to build the biggest X-ray observatory to date. So a little bit about me first. So my name is Luz. I am a software engineer in a company called Cosign. And ever since I was a kid I really liked space. I always had a really big passion for space. And yeah, eventually it was uh the industry in which I ended up which was a real pr pleasure and so far I've been at Cosine for over a little bit over five years now. Um I work there as a software engineer but I also do things like data science and data engineering. It's a little bit of a mix of
Speaker 2: All the three. So um that's what I currently do. And um so what is Cosign? Um Cosine is a company That is located in the Netherlands in a small town called Sossenheim. And we also have some technical facilities in Berlin, in Germany, and some offices in Benevento in Italy. We do various collaborations, think of small, large companies, could be anything, technical, scientific institutes, universities, but also bigger Companies with uh just like Issa or uh NASA or even uh the Japanese uh space agency, JAXA. So it's a bit all over the place, but mostly our focus is in the space industry
Speaker 2: So then what exactly do I do? I work in a team called High Energy Optics. And this is a business unit in which we develop and deliver all kinds of X-ray optics and systems. Our current biggest project is called the Athena X-ray observatory and it is part of the silicon poor optics technology. I will get more into detail about that later. Um within the silicon port optics we use uh wafers from the uh semiconductor industry. You can think of for example ASMR make uh ASML making chips. and other things for your phones or for your laptops. And we use these same wafers, but then for a different application, in this case, well, a big X-ray
Speaker 2: observatory. So this new Athena project, it's a big X-ray observatory that we contribute to as part of a big ESA mission. This is basically a really big project that um is not only part of just ESA and and Cosine, but Many different companies working it together. And I felt like this was very fitting because you know there's a lot of space missionists out there that are usually named after Greek gods or goddesses So I really couldn't pass up a uh on the opportunity to well talk about Athena and the city of Athens of course So yeah, like I said, New Athena. It officially stands for the New
Speaker 2: Advanced Telescope for High Energy Astrophysics. And um yeah, to this day it's the the biggest X-ray observatory today, so we try to keep this record. The focal uh length of the telescope will be about 12 meters, so you can imagine this will be a really really big telescope. The mission will basically a sharp focus on yeah the X-ray universe. We want to see things like neutron stars or black holes. Supernova explosions, active stars, all kinds of deep space objects that we relatively don't know much about yet And the planned launch is for uh 20-32nd, so we still have quite some time, but it goes fast, so
Speaker 2: and um we will be launching with the Ariane 6. 4 uh rockets. So I'm going to get a little bit into the uh rocket science part of things, so please bear with me. We have the following image, which basically describes how the X-rays will be caught into the detector. So X-rays don't work like visible lights. Um X-rays are of very high energy. And you can see this for example um if you if you imagine an X-ray as a rock and your mirror is water and you start to skip rocks, for example, if you just drop the rock just through the water, it will go through it immediately because
Speaker 2: well it's of such high energy that it will just goes straight through. So that's not really efficient if you want to actually capture these x-rays. So what we do instead of just dropping the rock on the water, we skip it And if we skip the rock, well, it will bounce off. So that happens with the x-rays as well. And then we can bounce off these x-rays to the detector and actually capture an image with it. So how do we do this? We have plates. We start with single plates. These are made out of silicon of the wafers that I mentioned. These are about the size of the palm of your hand, basically. And uh each plate has tiny ribs on there. Um so that means you can stack them on top of each other. And then you get something that's called a stack.
Speaker 2: When we have four of those different stacks, we can put them together and align them in a way that it actually reflects the x-rays. And then we can something that's called a mirror module. And we have to make, well, about 600 of these, and then we can put them in a mirror assembly, which is in front of the telescope near the solar panels. That is where the whole assembly will be placed. and will eventually reflect all the x-rays to the detector. So this is an example of a single plate. Like I said, it's about the size of your hand. And you can indeed see the little ribs that are in between And because if we stack these plates on top of each other, we get indeed this pore, yeah, this pore structure.
Speaker 2: So that's also why it is called silicon pore optics, and this is indeed the core technology Of this telescope. We have another image here which is uh this is a mirror module, and you can see that there's four of those stacks. So the stacks of plates again Two of our stacks, uh two of those stacks are uh aligned behind each other and they're aligned in such a way that they can bounce off the x-rays to the detector. Then we see here an example of someone who's working with the plates, and you kind of have an idea of the size of the plate as well. And we have to do all of this work in very, very clean rooms. And we call them clean rooms.
Speaker 2: And um yeah, this is very precise work. Of course, everything has to be aligned and and and cleaned and inspected so there's a lot of precise work going on there. So that's kind of the general idea of what this looks like in real life. So okay, so where's the software? Like we're just talking about rocket science, uh optics, cool, but there's no software yet. Well, the thing is within the company, software is not really like a deliverable for ESA or for any other company, but it's rather like an internal supports service. So we support the scientists and engineers within the company. We help them with, for example, things like data handling and the
Speaker 2: consists of data visualization, so think of looking into certain statistics, um data analysis, uh for example, see how the plates Their quality is over time, if they if they have a good or bad quality in the first place, and things like data processing with the help of data pipelines. So for example, if you have x-ray measurements Taken from Berlin, then can we see them? Can we read them? And how can we eventually visualize them as well? And last but not least, we also have data management, of course. So how do we keep track of all these objects? Each individual plate, where is it physically, what happened to it? Um and how do we store these data sets So that's actually where Django
Speaker 2: plays a big role in all of these different things. So it's kind of difficult to apply Django in a very traditional sense here because It is a scientific context and you need to be able to understand the data as much as this building the software around it. So this is for example um The case with flexible workflows. We need to make sure that we can adapt to development, but also to things like production or when we go towards production. So we need to be really flexible in terms of what can be done and what cannot be done in terms of how do we manage the data. We also need to keep in mind that RD and
Speaker 2: production can coexist together. They don't have to be separated in two different databases. But we just need to be careful that with one object you can do a lot and with one object the other object you cannot do so much And finally, if some of our data sets don't really fit into this pattern, then we decide to go for alternative data stores. So we store these data sets, for example, in MongoDB. This is the case with experimental data or one-time experiments that the scientists come up with. So our Django application, um what does this look like? So within the team we call this the uh SXRO
Speaker 2: database or silicon X-ray optics database Um this is a well-known term within the team. And uh in the back we use a MySQL database to store all the data in. And our main focus in this Django application is to track all the different components of Neo Athena and the different processes that we apply to these components. And of course, you cannot have a data management system without a little assistant. So this is our uh database assistant, Stacky. Um it was inspired by Microsoft Word's Clippy and well he's very loved within the team and uh started as an inside joke but now became a very big thing.
Speaker 2: So yeah, just to give an idea of what our data looks like, um we have attributes such as quality and status that we uh track of each individual component so think of the the quality of every single state uh plate sorry uh inspected or delivered or if a plate is stacked already Um and also whether the plate is of good optical quality or not, can we actually use it in our stacks or is it so bad that we have to just get rid of it? And of course, we also make use of components relationship, which is perfect for a relational database, of course This fits really well into the Django application as well. So here we track, for example, which blades
Speaker 2: is in which stack and which stack is again in which mirror module. And like I said before, we track the certain processes or as we call them tasks that will be applied to the different components So for example think of a plate inspection or the assembly of a mirror module. And finally we also track the history of all these different components. So this is kind of like our version control to see what happened over time. And to give you an idea of what this looks like, this is a screenshot from our Django app. It shows different tasks and processes and kind of like almost like a state machine type of thing Um we're not really into this stage yet, but we really do want to work towards a really strict
Speaker 2: state uh machine type of system. But for now we just track the different components and yeah apply all the different processes to them. And as you already can see in the screenshot we have used the default templates and almost the Django admin. like style for this uh for this app and that's actually because we we use the app min side we use it for our end users And of course the admin site is just to get an overview of the different components. We can apply certain CRUD operations. And it's used daily by the team. And of course there's kind of like a popular opinion against it. There's many people who say the admin is not for end users.
Speaker 2: Actually, the first book I read when I was just getting into Django, Two Scoops of Django, by Daniel Greenfeld and Audrey Roy. I would totally recommend this. Um it's it's even says that it's not for end users. So this is the first thing I read and then I was like, nope, still gonna use that use the admin because it just works. It just works perfectly for us. And why is this? Because we have a small user group. We have about 50 users. It's just a team, so no people from outside or other companies. And um they're thankfully very technically aware. I mean they're scientists, engineers. Most of them have PhD. So they understand
Speaker 2: data structures and the impact of changes when they do something, the database And also we looked into other alternative front ends. They were nice, but they kind of offered a similar functionality. So Why should we switch? We we were really happy with this system and it came out of the box so we just stick to it. We've been using it for a few years and we're still very happy with it So what does it look like? Um well everyone has seen a Django admin. This one is not so different. But we added some extra filtering and search capabilities. So you can see that for example on the top there's a little bit of extra filtering going on that we added
Speaker 2: That really help out the users. And we do this with the help of certain third-party packages. One of them being Django Simple History. That's the one that we use for the version control. We are a really big fan of this because we can track each individual stage stage for each individual component and it gives us a really good idea about what is happening within the workflows of our company. Then we also make use of Django QL. This is one that can actually be used in the search bar, and it is actually the Advanced queries for different kinds of query sets. If you want to look up certain nested relationships, for example, Django
Speaker 2: QL is the package to use. I really recommend it. And then we also have some smaller filtering packages which are used in the the right-hand side of the admin interface. And these just include things like ranges and lists and numeric sliders and drop-downs that can help us out make make it easier to find different components in the admin. So we also make use of the Django forms, which is a great thing by the way. We use this in a form we call we call this actually lab mode within the company. I don't know where the name came from, but it just stuck and it basically means any action that we perform in a lab or in a clean room So think indeed of the different measurements
Speaker 2: that are taken, the physical processes, mechanical processes, or even chemical processes. And the forms are really there to track the different kinds of processes and it is asked by the operators and scientists and engineers to track all of this manually. But of course you don't want to do this full time. You just wanna do your science. You wanna do something in the clean room and not just being busy with your administration. So We try to minimize the manual input as much as possible. Where we can, of course, we want to automate this and we try to do this a lot using our API, which I will get to soon. So one of the examples
Speaker 2: of a form that we have is, for example, the plate inspection. The plate inspection is a form that is used to detect particles or scratches or other type of defects on plates and operators use it with an iPad and a stylus and I just draw if they see anything that's wrong with the plates. Another example we have here of a form is one where we grade the quality of the plates in a plate container. So this is a more basic form of Django, but this works really well for us. And then we have a REST API. Like I said, some of the processes can be tracked automatically. So this is really convenient for us
Speaker 2: because, well, the REST API can uh be used for uh these types of process tracking. And we also use it as a sort of like an entry point for other tools and and data software and data tools within silicon port optics. So think of data analysis and synthesis tools. web -based data viewers where we plot and view all kinds of statistics. We use Jupyter notebooks in the form of Jupyter Hub Uh and we have the stacking robots and this is actually a really important point because we use the the stacking robots to stack the plates automatically. This is a full-on automated process And yeah, the API is used to track this whole process
Speaker 2: so each plate gets us updated in real time as we stack the stack. And finally, we have the observability tools. Think of things like Apache Airflow and Grafana to make sure that everything runs the way we want to and at all times. Some of the impact uh important packages that we use for the REST API, um, for example, is Django REST framework, just for to make the REST API work in the first place. This works very well for us. It's very consistent and has been working for years and will continue to use it. And then we have uh DR Spectacular, which is one of the packages that we use to generate different schemas and documentation, of course, with the Open API standard.
Speaker 2: So this is very useful for us. And lastly, we have Django filter. And this we use very intensively to filter on certain query sets using API query parameters in URL. So we don't physically send Django up to space, but it will definitely help us get there, so that's really nice. Um yeah, Django is really much the software backbone for this project, New Athena. It is a general purpose framework, but it works really well in the scientific content. context at least for us. It's very flexible. It can it's really adaptable to different kinds of environments. Think of RD but also uh eventually production
Speaker 2: when we actually have to build these mirrors for Athena. And yeah, we of course make sure uh to use the the features that come out of the box, so the admin side, um which is the primary interface for this app. We have the forms to track the different components and their processes and our statuses. And last but not least, the API to handle the data. This could be either through visualization or data processing or even observability, which gives us a really good idea of what is happening within our app. So, what is next? Um we will ramp up uh towards a full-scale production.
Speaker 2: by the end of next year and then we will actually make the mirror modules that will go into the telescope. So we're kind of like tr yeah, going through this phase where we're trying to set everything up and make sure that everything will work for production. So that also means that we have to set the software up in such a way that it's handle uh able to handle this full-scale production environment. So for software that means we're looking into well more state machine driven workflows for the processes and the different components and of course the tasks that we track And we want to look into a clear separation between R and D in production because at this point we feel like the the projects and the experiments that we do
Speaker 2: will divide in such a way that we would need a clear separation, maybe even in different databases. And then finally, well, we hope to launch the telescope in uh 2037. So I hope to see you there in the live stream. And that's it. Thank you.
Speaker 1: I'm sure there are questions. And I'll start here.
Speaker 3: Yes, in my experience when scientists start building software it either goes magnificently like accidentally inventing the World Wide Web But they also make like really embarrassing mistakes. So are your scientists and your software team like separated or do they work Together. Do they make embarrassing mistakes?
Speaker 2: They're indeed separated. Um so I work on the software team and of course we just have scientists that focus on well really inventing things and working in the clean room for example and just doing R<unk>D stuff related to only the mirror modules and the hardware that is involved. Um so they usually come up to me if they need something like a new software feature or um this could also be some data analysis that they need. So it's not only a feature in Django, but it could just be anything. And that's also why My work is kind of like a mix between all of these things. So uh yeah, it's a bit hybrid, but uh we do have a clear separation.
Speaker 4: Um by much by how much more will the production scale up? Currently you're doing like certain plates, how much are intended to be done?
Speaker 2: Wait, sorry. Oh there you are.
Speaker 4: No worries. Have you got my question?
Speaker 2: No, sorry.
Speaker 4: By how much um the production will scale? You're doing like currently a few plates a month? How much more will it be done?
Speaker 2: We do currently about two stacks a day And um yeah, it's already quite a lot, but eventually when we need to ramp up, we'll probably have well certain kind of like milestones that we need to work towards. And then We probably have to stack day and night throughout the weekends and everything and now on Friday afternoon the clean room is just s silent. So yeah, at some point we need to ramp up in that sense as well.
Speaker 1: In the meantime I go here, we'll run around.
Speaker 5: Thank you very much. That was a really great talk. It's great to see how you use Django in such a unique way. I was just wondering very quickly about the hosting story. So do you is it in the cloud? Is it on-prem or or or what do you guys do?
Speaker 2: So everything uh that we have in terms of software is all hosted internally. Um and this is also required by some of the space agencies. They don't want us to work with the cloud, so Um we have a lot of servers that that are not very common in that sense. Um we make use of a lot of open source packages and and for example we have our own um mail server as well, so it is very Very secured.
Speaker 6: But uh how redundant is it? Like I'm assuming you're not running manage. py run server uh and just having that like uh if you lost the data and all of that like how would it affect the project and How how much redundancy do you have on that site?
Speaker 2: Well we next to just our software team we also have an IT department that makes sure that we have proper backups that they host everything well internally so this is something they take care of but it's not just run server and hope for the best thankfully
Speaker 1: More questions? There's one here.
Speaker 7: Thank you very much. Because there's a lot of people working on Django in the room. Is there anything you wish Django did already that would have helped this project?
Speaker 2: Well that's a good question.
Speaker 7: And and that might be something people work on during the sprints.
Speaker 2: Well I think it would be interesting to explore explore the the scientific side of Django and see where we can use it more in a scientific context because I feel feel like every time I talk to people within the space industry, they don't have a system like this at all. They they don't track it as well. And I think Django could be of great use within this industry in that sense, just in general
Speaker 8: This is a subset of the previous question. Um but when you were doing forms and like your own workflows in the Django admin, was there anything specific that you Uh could have wished that the Django admin supported or how did you make those uh forms show up? Uh
Speaker 2: so for the forms we use a lot of uh custom business logic to make sure that each procedure goes as it should it it gets a certain state once that procedure is done. Um so there's a lot of custom business logic that we build in obviously. In that sense, we don't make use of external packages that also handle this logic. But yeah, that would be something maybe that could help us more if there's some standardized Django way for that
Speaker 7: Thank you, Luz, for such an interesting talk. I assume that changing some lines of Django is much quicker and less expensive. uh than changing a robot that has to emit all these uh parameters. So I assume that when you get some new equipment then it's the Django that has to change to match
Speaker 2: yes
Speaker 7: that and then s you have to do a lot of work to Do that. Is that something that happens often or hardly ever or
Speaker 2: not? Especially with the stacking robots, we we have a bunch of them and They just do the work and that's all very stable. We don't plan to buy a lot more, thankfully. If we do, it's probably going to be of the same series. So That's covered. Um and uh we have a software package, uh wrote it ourselves that is outside of Django but does also the handling of this hardware as well. So um because we want to just also have this internally so we have full control over it and it's also easier then for us to apply this back into Django. Um so yes it takes some more time to implement all of these custom things, but at the same time if we were to outsource it it would probably cost us even more. So and now we have full control over it so we can decide the way we want to if For example, we have
Speaker 2: a uh experiment going on for maybe like a week or two, then we can set the these settings ourselves without having to go to another company and ask them, hey, can you please do this for us So it kind of gives us more freedom in a way.
Speaker 1: Thank you, Luz. One last question from my side. Where are flying the observatory to
Speaker 2: Where?
Speaker 1: Yeah.
Speaker 2: I don't know yet actually. Maybe between L1 and L two, something like that.
Speaker 1: Okay, Lagrange points. Yeah. Or we'll be watching 37, I believe.
Speaker 2: Yes.
Speaker 1: Thank you a lot. Please give a warm hand.
Speaker 2: Thank you.
April 15 - 17, 2026
Athens, Greece
Note: We understand that names change, people change, and bodies change. We respect each individual's journey and privacy. If you have any concerns about a video or need us to remove content, please don't hesitate to contact us. We will handle your request with care and promptly address any issues.
Published October 26, 2026
Published October 26, 2026
Published October 26, 2026
Published October 26, 2026
Published October 26, 2026
Published October 26, 2026