Inside the life of an AI engineer

What position do you hold in the company and how did you get there?

I am an AI software engineer at AITAD. After my studies, I looked for a company where my favorite study subjects were part of the working day, and that's how I ended up here.

What does this job involve?

In my role, I focus broadly on collecting data and making it usable for our future model. To do this, I look for anomalies in the data pool and try to find the appropriate model that fits our algorithm and the desired functions.

What activities and tasks does this position include?

Basically, my work consists of three interdependent areas:

• First, I participate in sensor evaluation. This means that I select the appropriate sensors for the project from our existing sensor portfolio.
• In data collection, I collect data using the previously selected sensors and make it usable for the company and the project.
• As part of algorithm optimization, I adjust the models and try to improve the results.

What does your working day look like?

Usually, this starts at eight o'clock with checking the model training from the previous night. Then, our software team discusses the current projects and tasks. After that, I focus on my main tasks (see 3). In the evening, there is an informal exchange, and I start the model training for the night.

How did you find the job? What did you study or what qualifications do you have, and why did you choose them?

I attended and successfully completed the digital signal processing course at the University of Strasbourg. Subsequently, I also applied in Germany because I live near the border and wanted to leave my comfort zone. The subject area of AITAD and the advertised position exactly matched my interests and search criteria.

What qualities do you need for the job?

Basically, you need knowledge in the areas of artificial intelligence (AI) and signal processing. A mathematical background or interest in mathematics is also advantageous.

In daily work, you need to be agile and adaptable to meet the respective requirements of many different projects. Communication skills are important for teamwork.

What role does collaboration and exchange with colleagues and other professionals play in your work environment?

For us, internal and cross-team exchange is of great importance. We offer our customers both hardware and software components and therefore need to exchange information about the respective requirements.

If I don't speak with the hardware team, I don't know how to adjust the software to make it fully functional on the existing hardware.

How well does the degree prepare you for working life?

The degree has prepared me well with the theoretical foundations for my current areas of responsibility.

However, you are also confronted with challenges that you did not know from university life. These include, among other things, delayed feedback from customers or service providers, which can delay the project, or "unclean" data that must first be adjusted or newly recorded.

Which technologies or trends in electronics currently excite you the most?

At the moment, I am very enthusiastic about the Python programming language. I use it extensively in my daily work and benefit from the large user community and the extensive documentation. Many problems I encounter have already been faced by other developers, so I always find suitable solutions, learn something new, and can continuously improve my and our solutions' performance.

What challenges do you see in the current electronics development landscape?

Currently (and probably in the future), we are struggling with the performance-price ratio. Our customers want maximum performance for the lowest possible price. From economics, we know that it is difficult to achieve maximum yield with minimal effort; there is no mini-max principle - this also applies to the development of AI solutions.

We must therefore be able to convince the customer that they will receive better performance if they invest more money in better sensors. For this, we must also guarantee better function and performance.

What does the future of electronics development look like in the context of AI and machine learning?

I assume that in the future we will see a strong increase in the use of AI, especially in the industrial environment. There will be more and more hardware and software variations, which will make the areas of application and possibilities differ greatly from each other. Associated with this, AI models and algorithms will improve and evolve.

Furthermore, I assume that the systems will become more energy-efficient and possibly even be operated with their own heat.

Can you tell us about a particularly exciting innovation in the electronics industry that you have personally experienced?

I recently contributed to the development of a robust voice control system. We managed to ensure that spoken commands were recognized clearly and distinctly by the application, despite loud background noise. For this, we worked with beamforming technology, which allowed us to separate the actual command from the background noise using two microphones.

What impact does AI have on daily work?

AI is my daily work. By collecting data and creating and improving models, I actively contribute to solving problems and ensure that the use of artificial intelligence continues to increase.

In your opinion, how does electronics development contribute to solving societal challenges?

The development of electronics and especially AI helps to relieve working people. It can help automate repetitive tasks, giving people time for creative work. It can also be used in dangerous situations to prevent harm.

This not only prevents unnecessary errors, but also achieves economic and ecological goals, as intelligent condition monitoring can optimize the lifespan of technical components.

What highlights does the job bring?

I am fascinated by discovering new and previously unknown patterns in datasets. How does variable X correlate with variable Y and what happens when there are small deviations from the norm - can I detect them?

In some cases, we even discover additional dependencies and influencing factors that we were not primarily looking for and did not know about.

Are there any projects that were a particular challenge or that have had a lasting impact on you?

Right at the beginning of my work, we were involved in a large project where we monitored and assessed the condition of engines in production. We collected data on site and had a lot of 'raw' data. These were difficult to use and demanded everything from me at that moment. However, because I was able to master this challenge, I am now able to approach current and future challenges more calmly and relaxed.

Are there any electronics-related projects or hobbies outside of work that you particularly enjoy?

Currently, I am developing my own powerful PC workstation. I am assembling it from (partially used) individual parts and trying to achieve maximum performance this way. Once the project is completed, I will also be able to work on AI models at home.

What should aspiring electronics developers keep in mind?

You need to think 'out of the box', as they say. Don't focus too much on the raw data. You need to know how it is structured and what other factors influence it. In our case, for example, I need to know the machines from which the data originates and understand how they work. Only then can I find the appropriate solution path, starting with the selection of sensors, for the project.