Database Reference
In-Depth Information
In regards to conferences, papers, and blogs, there are so many! These days,
if you build a community around yourself, the news and people start to find
you. In 2013 we did two challenges—one was with Kaggle and the other one
was our own self-hosted challenge. With Kaggle, we created a challenge called
“Create an algorithm to detect North Atlantic right whale calls from audio
recordings, to prevent collisions with shipping traffic.” The Kaggle data sci-
ence community improved the state of the art in existing bioacoustics mod-
els in just two weeks, a significant achievement. Thanks to this, whale voices
in an economically and environmentally valuable area are now detected at
a much more accurate level. This has inspired oceanographers to actually
change the methods they use to analyze audio. Nowadays, there are machine
learning-focused tracks in bioacoustics. For our own self-hosted challenge,
the Marinexplore [Planet OS's previous name] Earth Day Data Challenge, we
invited people to work with data we had and to share ideas on how to analyze
and how use that analysis of the ocean in different businesses. The winner of
the challenge looked at growing algae near the coastline of Brazil.
We've learned the most through active interaction with professionals, both in
challenges like those I just described and as we've grown our product. We've
cycled through more than 30 different releases over the two-year period.
From each release we've learned something new. We make blog posts and we
present and interact at conferences like Strata or the Society of Exploration
Geophysicists or the American Geophysical Union.
We've also learned from the people we've hired. We've hired some oceanog-
raphers from the community to work with us, and they have taught us every-
thing known to date about the physics of the ocean and the dynamics of the
ocean. Roberto De Almeida from the Brazilian Space Institute worked with
us a great deal, and we as a company learned a lot about the ocean from him.
He is also the author of Pydap, the Python implementation of the OPeNDAP
data exchange protocol, so that helped as well. We've also learned quite a bit
from John Graybeal, a US cyberinfrastructure expert. We've also worked with
Chris Clark, who was leading a large research group in bioacoustics. We've
also learned a lot from the friends and colleagues of these experts.
Lastly, of course, we've learned from our customers. They are in the ocean
daily, so they know best what's going on. Our customers teach us about what's
relevant and which ocean parameters to focus on, such as wind-wave ocean
currents, bathymetric ice, and all the other crucial parameters affecting their
daily activities.
Gutierrez:
What does a typical day at work look like for you?
Karpištšenko:
We are a company that operates 24 hours a day and 7 days a
week. We have offices in Sunnyvale, California, and in Tallinn, Estonia. I mostly
work in Estonia, so my day begins by opening a laptop and checking what has
happened during the night and on the other side of the planet. The day ends in
