INTERVIEW

While hunting for a summer internship, I had two fundamental criteria in mind. Firstly, I wanted to work on deep learning. Secondly, I really wanted to intern somewhere in Japan. Machine learning in general and deep learning in particular, have been the focus of my self-initiated learning efforts for the past year as a sophomore. Through MOOCs, personal projects, voluntary research assistantships and internships, I have built a sizeable base of knowledge on these subjects which weren't a part of my coursework this far. To enhance my skills further, I felt that receiving feedback on my ways of perceiving a data science problem and solving it by means of deep learning was a necessity. This led me to focus my internship search on finding a role centered on deep learning. On top of the internship's project theme, the fact that it was in Japan was a great determining factor for me along with its short- and long-term benefits. Having studied Japanese in Osaka University for a year on MEXT Scholarship before, I was looking forward to reconnecting with Japan and Japanese. I also wanted to experience the life as an R&D engineer in Japan. Satisfying both, the data analytics internship opening in SEI was the perfect match for me.

During the internship, I implemented a deep learning model capable of segmenting satellite images according to a given set of object classes (roads, buildings, vegetation etc.). In other words, this corresponds to building an artificial neural network which can learn to mark the mentioned classes on any satellite image after training on a set of labeled satellite images. Opting for a neural network architecture called 'U-net', widely used in image segmentation, I progressed through cycles of training, validation and fine-tuning to enhance my model's predictive accuracy. After obtaining the final results from the test data, I presented my work to data analytics department manager, engineers from my team and HR staff.

This one may feel a little odd but thanks to the “communal stretching time” of the R&D office, I found out that a few minutes of post-lunch break stretching was actually capable of giving one the afternoon boost you might need. While at first I was slightly skeptical about the rounds of hip twists and shoulder squeezes dutifully performed by the majority of my co-workers, I have really come to appreciate the fact that awakening and taking care of your body work wonders for your productivity. I now feel like I have built a powerful habit which is applicable to a wide range of contexts demanding mental engagement.

First and foremost, the internship solidified my motivation to work on data analytics. I really enjoyed working on a machine learning project theme, especially on one that introduced me to the field of computer vision. These will, for sure, have lasting influence on my career focus and research interests in the future. Other than what I actually did as a part of my internship project, the opportunity of participating in the R&D environment of a world class company steered my expectations from a career in corporate R&D towards a more realistic standing. It helped me better judge what benefits there might be for me in a career of this field. Furthermore, the internship was my first experience working abroad. The ease of communication with everyone in the company and the support given to interns on all sorts of problems we faced left a positive impression. I had been thinking about working or studying abroad after graduation and my lovely first experience in SEI has definitely encouraged me to do so.

You should be keen in discerning what kind of projects you would like to work on if you decide to join the company for a full-time position. Your internship project will be a good indicator. However, the projects that other team members, especially your mentor, work on are also good measures of both the scale and complexity you might encounter in your research. Be enthusiastic in interact with your team mates to understand better the work that they do. Your daily routine may not require you to collaborate with people other than your mentor but everyone at SEI is eager to provide guidance and give detailed explanations when asked.