Satellite engineering has changed more in the last decade than any other part of aerospace, and the change was industrial rather than scientific.
Producing at rate, not producing once
Building one exquisite satellite and building hundreds of adequate ones are different engineering problems, and the second one is now common.
Four things change when you produce at rate.
Design for manufacture becomes decisive. A design that is elegant but slow to assemble caps the production rate regardless of how good it is.
Supply chain becomes the constraint. Hundreds of units need hundreds of every component, qualified, on schedule. Qualifying second sources becomes a standing activity rather than an exception.
Test has to be fast. Traditional environmental test campaigns take weeks per unit. At rate, test must be sampled, automated or shortened, and deciding what can safely be reduced is a serious engineering judgement.
Reliability becomes statistical. Individual unit failures are tolerable if the constellation delivers service. This is uncomfortable for engineers trained to eliminate every failure mode and it is the correct approach for the mission.
The skill that limits a constellation programme is production engineering, and it is usually recruited last.
What the direction covers
The scope: satellite systems and payloads, communications, remote sensing and geospatial data.
Four areas.
Satellite platforms. The bus: power, attitude control, propulsion, thermal, command and data handling.
Communications payloads. Antennas, transponders, link budgets, modulation and the spectrum question below.
Remote sensing payloads. Optical, radar, hyperspectral and infrared instruments, and what each can actually measure.
Ground segment and data. Covered below, because it is where programmes come unstuck.
The ground segment nobody budgets for
The most consistently underestimated part of any satellite programme.
A satellite in orbit is useless without the infrastructure to talk to it, task it, receive its data and turn that data into something usable.
Ground stations. Antennas, radio equipment and the sites to put them on. Geography matters, because a satellite in low orbit is only visible from a given station for minutes at a time.
Scheduling and tasking. Deciding what the satellite does and when, subject to power, thermal, memory and contact constraints. A genuine optimisation problem that grows harder with fleet size.
Data downlink and handling. Modern payloads generate more data than can be downlinked, which makes onboard processing and prioritisation a design requirement rather than an option.
Operations. Monitoring health, responding to anomalies, performing manoeuvres and collision avoidance. People, on shift, indefinitely.
Cybersecurity. Command links are a control channel for physical assets, which makes them a security problem with the same seriousness as any industrial control system.
For a workforce plan this is the practical correction: a satellite programme employs more people on the ground than in building the spacecraft, and ground segment roles are the easiest to fill from adjacent industries.
Where this sits in the domain
Satellite engineering and Earth observation is the fourth of nine directions in Astra Trainer's space, aerospace and new mobility domain, sitting on astronautics and space systems and connecting to the space economy, policy and law direction where spectrum, licensing and market structure are covered.
The ground and data half draws on the AI, data and computing domain, particularly cloud computing and DevOps, data science and analytics, and cybersecurity, and the communications half draws on semiconductors and electronics for RF and wireless. Partners running satellite services usually need all three. You can see the nine directions here.
Earth observation is a data business
The commercial reality of remote sensing, stated plainly.
Capturing an image of the ground is the input. The value is created afterwards, and by different people.
Processing. Raw sensor data requires radiometric and geometric correction, atmospheric correction, orthorectification and calibration before it means anything. This is specialist work and it is where accuracy is won or lost.
Analysis. Turning corrected imagery into information: change detection, classification, measurement, time series.
Domain interpretation. Deciding what the information means for agriculture, insurance, forestry, defence, maritime or infrastructure. This requires knowing the application, not the satellite, and it is the same point made in the drone article.
Delivery. Getting the answer into a customer's workflow in a form their systems accept.
Three consequences for hiring.
Most of the workforce in an Earth observation company is not aerospace. It is geospatial analysts, data engineers, software engineers and domain specialists.
Owning satellites is not required to build a business in this, since imagery can be purchased, which is why many of the companies in the sector operate no spacecraft at all.
And the scarcest people are those who understand both the physics of what the sensor measured and the domain the customer works in. That combination is rare and it is the same two-halves problem that appears throughout this section.
The roles, named
Satellite systems engineers.
Production and manufacturing engineers for constellation assembly. The volume-hiring role.
RF and communications payload engineers. Persistently short.
Remote sensing instrument engineers. Optical and radar.
Ground segment engineers. Antennas, networks and station infrastructure.
Satellite operations engineers and controllers. Shift-based, growing with fleet size.
Mission planning and scheduling specialists.
Geospatial analysts and data engineers. The largest population in the downstream industry.
Spectrum and regulatory specialists. Small, slow-moving and genuinely gating.
Who can be trained into it
Manufacturing engineers and production technicians. From automotive, electronics or medical devices, into constellation production. They bring rate, yield and supply chain discipline that traditional space programmes never needed.
Telecoms engineers. RF, link budgets, modulation and network operations transfer directly to communications payloads and ground stations, and this pool is large.
GIS and remote sensing analysts. Already in the downstream business, and a short step into satellite data roles.
Network and data centre engineers. Into ground segment and data handling, where the work resembles distributed infrastructure.
Cybersecurity professionals. Into command link and ground system security, which is an underserved requirement.
Broadcast engineers. An unusual conversion and a strong one, since satellite broadcast experience covers RF, ground stations and link management.
Spectrum, licensing and export control. Operating a satellite requires regulatory authorisation and spectrum assignment, which involves national regulators and international coordination, and lead times are long enough to gate a programme. Satellite technology and technical data are subject to export control in most jurisdictions, and remote sensing is separately licensed in several. Training builds engineering understanding. It does not confer any licence, spectrum assignment, operating authorisation or export clearance, and regulatory strategy requires qualified specialists engaged early.
What to take from this
Constellations made satellite building a production-rate problem, and production engineering is usually recruited last.
The ground segment employs more people than spacecraft manufacture and is the part most often under-budgeted.
Earth observation value is created in processing, analysis and domain interpretation, and most of that workforce is not aerospace.
Spectrum and licensing lead times can gate a programme, so those specialists belong in the plan early rather than late.
And telecoms engineers, manufacturing staff and broadcast engineers convert better than aerospace graduates for most of what is actually open.
What changed with satellite constellations?
Production rate became the constraint. Design for manufacture, supply chain, fast test and statistical rather than absolute reliability all became central, which needs manufacturing people rather than more spacecraft designers.
What gets underestimated most?
The ground segment. Ground stations, scheduling, data downlink, operations and cybersecurity employ more people than building the spacecraft, and are routinely under-budgeted.
Where is the value in Earth observation?
In processing, analysis and domain interpretation rather than in capturing the image. Most of the workforce is geospatial, data and software, and many companies operate no satellites at all.
Who converts into satellite roles?
Telecoms engineers into payloads and ground stations, manufacturing engineers into constellation production, GIS analysts into downstream data, and broadcast engineers, whose RF and ground station experience transfers unusually well.
Where does this fit in the domain?
Fourth of nine directions in Astra Trainer's space, aerospace and new mobility domain, drawing on AI and computing for the data half. You can see them here.