One of our favorite things to do at Megagon Labs is welcome interns from all over the world into our office. This spring, we had the pleasure of working with two talented interns, Seiji Maekawa and MeiXing Dong. They worked on interesting research such as active learning and information extraction. Below they share their experience and the lessons they learned along the way.
Seiji Maekawa
University: Osaka University, Japan
Program: PhD in Information Science and Technology
Project: Low-budget active learning. Aiming to reduce labeling costs (i.e., human effort) by focusing on informative data samples to train language models. (Project team: Dan Zhang, Hannah Kim, and Sajjadur Rahman)
What was your experience working with the team at Megagon?
They made me feel comfortable and like part of the team. We had discussions outside of meetings on a more personal level, making it a more relaxed setting. On the work level, they helped me see how researchers work in a company. I received clear guidance which helped me understand what I had to do. Our team members always reminded me how our project connected to the overall picture, bringing me clarity and motivation. I appreciated that the use cases are more concrete for companies, in contrast to my experience in academia.
Overall, I had a great experience collaborating with everyone in the office, especially during our conversations. The discussions over lunch and after presentations and meetings really helped to broaden my research knowledge. I was able to take input and have discussions with everyone, bringing new blood to our project.
How did you benefit from the internship?
This internship gave me an opportunity to generalize my research skills to another data domain, natural languages, which is different from the main topic of my Ph.D. work.
NLP and Graph Management are in the same area, but I wanted to fine-tune my skills in NLP because machine learning models for natural languages are treated differently based on the dataset. Megagon’s focus on machine learning is different from my research, so I wanted to dive into it.
I gained real-world experience working alongside a great team. I came wanting to see how researchers in a company bring their research to products and stakeholders, and how research is useful to businesses. Thanks to their help, I got more used to the NLP implementation, saw projects come to life, and gained new skills.
Coming from Japan, what did you enjoy about the Bay Area?
I enjoyed the great weather, conversations with open-minded people, and beautiful sceneries. One of the most impressive things is a variety of people who have different backgrounds. I found that people can naturally respect each other by recognizing the differences of their backgrounds.
I also had fun exploring San Francisco. I visited a number of local breweries and spots with ocean views. They were totally great experiences for me.
Meixing Dong
University: University of Michigan, USA
Program: PhD in Computer Science with a focus on NLP and Computational Social Science
Project: Improving factual information extraction from language models such as BERT or T5. Specifically, exploring how we can add contextual information to the querying process. (Project team: Nikita Bhutani, Hayate Iso, and Estevam Hruschka)
What did you want to achieve during your internship before you started?
What's something you learned during your internship that you'll take with you?
I learned how there can be more iteration during a project than expected, even despite lots of prior planning, and that it’s okay to just see where experiments go and revise plans accordingly.
What was your experience working with the team at Megagon?
Coming from Michigan, what did you enjoy most about the Bay Area?
I love how there’s such a variety of activities to do, from good eats and concerts in the cities to wildlife viewing and hiking in the mountains and along the coast. Having consistent sunshine is a great mood booster too.