Can you share a bit about your background and what led you to work in the field of data science?
During the pandemic, as remote work became the norm, it became challenging to develop hardware projects given my academic background in Microelectronics. Consequently, I transitioned into data science. Expanding my knowledge in this field through personal projects. Additionally, academic projects provided valuable insights into data science, sparking my passion for the field. My experience further grew when I joined Ubotica as an intern in 2021. Where I leveraged my engineering background to work on a data science project. What I appreciate about Ubotica is the opportunity to blend theoretical concepts with real-world applications. This allows me to seamlessly integrate my hardware background to develop cutting-edge projects, particularly in Edge AI.
What is your role at Ubotica, and what does a typical day look like for you?
Mainly focusing on data science, I leverage data to solve complex business challenges. A typical day would be collaboration between teams, we communicate what we are working on, and any issues that have arisen and we discuss possible solutions. I spend a huge amount of time analysing data and researching solutions.
What specific projects are you working on now and what are their challenges?
Projects: 3CSD A project that integrates AI algorithms to run on edge devices, involved in serial use cases. Some use cases require additional knowledge. It is important to develop specific expertise for each use case. We often need to understand concepts from other fields. This necessitates extensive research and review of scholarly papers. While everything might initially seem to work well, unforeseen limitations can arise, requiring further investigation and problem-solving. Additionally, the challenge of meeting real-time requirements pushes us to carefully design lightweight architectures that balance high accuracy with minimal resource consumption.
How do you approach solving complex problems in this project?
When solving complex problems, I generally start by breaking down the issue or the task to be done into smaller, more manageable parts. I gather as much relevant information as possible, often from previous studies, research papers as well as my colleagues. Next, I brainstorm potential solutions, evaluating each one for feasibility and potential impact. I then implement the most promising solutions iteratively, testing them to identify any limitations. When these limitations surface, I reassess and either modify the existing approach or develop a new solution to meet our requirements. Throughout the process, I stay flexible, ready to adapt if new challenges or insights emerge.
Can you explain the role of AI in transforming traditional Earth Observation methods?
AI is playing a transformative role in traditional Earth Observation (EO) methods by significantly enhancing data processing, analysis, and interpretation. Traditionally, EO relied on manual or semi-automated techniques to analyse satellite and aerial imagery, which could be time-consuming and less precise. AI automated these processes and offered more advanced and rapid methods to achieve promising results in a short time.
What types of data do you work with, and how do you ensure its quality and accuracy?
We utilise various data types such as images and hyperspectral data, and then apply advanced algorithms, using machine learning models, to process this data. Additionally, we often use machine learning algorithms for preprocessing, which helps us achieve higher-resolution data.
How important are collaborations, like our partnership with IBM and Open Cosmos, in achieving our goals?
I believe this partnership enables us to, leverage our shared goals as we align and have the same vision, this guides us to solve many complex space challenges. In the same way, we achieve goals with CogniSAT-6, we can enhance AI processing performance onboard using various technologies from different companies.
What advancements or innovations are you most excited about in the field of EO and AI?
Yeah. Honestly, I am particularly excited about new satellite technology, specifically, because, as data scientists, the first thing that we start thinking about is the data and how quality the data is. If we are able to improve this technology to get higher resolution imaging and be able to capture enough data that will improve our results it’s very exciting for us. And of course, this kind of improvement allows us to monitor Earth and develop more advanced solutions for different Earth observation missions. With the availability of data, we will be able to develop more and more AI solution based algorithms that allow us to detect patterns and provide some kind of prediction, classification, and detection tasks. So I believe that there is a lot to do, specifically, with the availability of data and the data that we have at hand. Also with improved hardware available we can achieve more advanced results in terms of Earth observation we can use AI to handle issues.
Where do you see the future of Live Earth Intelligence and Ubotica’s role in it?
Looking ahead, I believe that Earth Intelligence Services will play a crucial role in the future of Earth observation and predictive tasks. Ubotica already has, SPACE:AI technology, and, I believe that we will continue to advance Live Earth Intelligence Technology. We still aim to develop cutting-edge solutions to integrate both AI and Earth observation. The main aim is to help solve some issues such as climate change or disaster management with Earth observation. I believe SPACE:AI is the key for us as the first step towards improving Earth Observation and to be able to achieve more with this kind of technology in order to specifically limit some issues and monitor Earth accurately.
What do you find most rewarding about your work in data science and AI?
It is an opportunity to contribute more, to the advanced technology and specifically it’s exciting to work on real-world issues including some issues that we can offer others for their Earth observation missions. I believe this kind of mission push us to understand also what is behind in terms of science doing more research and how to implement more technical skills to develop real-world solutions. We enjoy focusing on research and engineering to develop innovative solutions for tackling complex challenges.
Do you have any advice for aspiring data scientists who are interested in working on impactful projects like ours?
Definitely, the first thing that they should consider is to master some programming language like Python, study machine learning fundamentals, stay curious and keep learning, and seek out opportunities to work on real-world problems to gain practical experience. For instance, in the space field, undertaking various projects and internships allows individuals to enhance their skills and gain valuable insights from experienced staff.