Employee Spotlight: Shaunak De, Data Scientist
In our mission to help clients understand what is happening on and to the Earth, we’ve created a culture that pushes us forward to deliver exceptional results. Orbital Insight created its Employee spotlight to illustrate the culture and people needed to support our mission.
In this month’s Employee Spotlight, we interview one of the many exceptional team members at Orbital Insight that is changing how businesses, governments, and NGO’s understand the world.
Q: Data science has become a buzzword in technology. How do you know, as someone in the field, how to pinpoint companies that are truly innovative and are doing interesting work with big data, versus those that just claim to?
A: The data science industry, and the problems it’s trying to solve is very unstructured at the moment. I view it as the aviation industry when the Sopwith Camel was the fastest aircraft in the sky. It is an exciting time filled with innovation and fast-paced progress — but unfortunately, sometimes semantics and lexicon become difficult to assign in the process. Personally, I revert to the first principles and look at the 4 V’s of data analysis: volume, variety, velocity, and veracity. It boils down to one simple question: Are you handling large volumes of fast-moving diverse datasets, all of which, most importantly create value to someone in the world.
Q: What projects are you currently working on?
A: I am currently working on our oil inventories algorithms where we are in the process of fusing information from Imaging and SAR satellites, along with ancillary sources of data like geolocation, expert domain knowledge etc. Some of the biggest challenges have come from integrating expert knowledge into the product. Specifically, we were trying to mark crude and non-crude tanks — and apart from tank metadata we also incorporated the opinion of people who had been in the oil and natural gas industry for a long time. However, on several tanks — they did disagree with each other. I developed a Bayesian inference model to incorporate “suggestions” (if you will) from these experts, indications and from the metadata we had from the tanks to come up with a probabilistic prediction of if the tank was crude or not.
Q: What are your passions outside of Orbital Insight?
A: I believe that the driving force of the universe is economics, and the truth of the world can be traced by asking simple questions around who paid whom, how much, and why. In my free time I love exploring connections in the world, and people’s motivation in acting how they do.
Recently, I looked at some data from a lifestyle magazine to see if patterns would emerge in its Instagram followers, what they “liked” and why they liked it. (Apologies for the terrible pun). I found that it turns out there is a very distinct pattern behind posts that did well, against those that did not. You can learn more about it here.
I also looked to see if I could simulate the SAR returns from an oil-tank with a Point of View Rendering engine (something similar to a 3D graphics engine) to understand the nature of SAR scatter.
Another project that I recently worked on was the detect point targets like oil-tanks and ships in UAVSAR data — using low-end computational techniques that would work on my home laptop involving the use of SAR polarimetry.
I am also building my own backyard radar that will be mounted on a bicycle and use cut-up baked beans tins as antennae. This project is approximately 40% complete. This is what the completed system could look like:
At Orbital Insight, we like to challenge ourselves and have fun! We know what we do is intense and that’s how we choose to live. Whether it’s an internal hackathon or team go-kart racing, we find that our performance is directly linked to our sense of satisfaction with our professional and personal relationships.
To learn more about career opportunities at Orbital Insight email email@example.com.