"Building the robot" is one job out of a dozen
Ask most people what a career in robotics looks like and they picture one person: someone in a lab building a physical robot from scratch. In reality, almost no one does that alone, and the industry has far more entry points than that image suggests. On the hardware side alone you have mechanical engineers designing structures, linkages, and drivetrains; electrical engineers building power systems, sensor wiring, and PCBs; and embedded engineers writing the firmware that makes motors, encoders, and sensors actually respond in real time. These are three genuinely different skill sets, and most robotics companies need all three, not one generalist who does all of them adequately.
Then there is the software and AI side, which has grown enormously and now often outweighs hardware headcount at many companies. Perception engineers turn camera, lidar, and sensor data into a model of the world. Planning and controls engineers decide what the robot should do and how it should move, from classical control theory to model-predictive control. Increasingly, teams also build and fine-tune foundation models for manipulation and navigation, which has opened a real path in for people with strong machine learning backgrounds who have never touched a motor. None of these people build the robot in the traditional sense, and all of them are, professionally, in robotics.