Talking story, grad school memories edition

Diving to the bottom

I was going back through some of the materials that I’ve put together for various and sundry job applications, and came across this piece. It’s not perfect, but still a good story, so I figured I’d throw it up on the blog. Many of the challenges I encountered as a grad student are so relevant to working in industry it is always entertaining to think about the parallels. I might write this differently today, as I’ve done some very neat (and innovative) stuff since I wrote this last year, but leaving it here as a reminder to myself.

Q: What is the most innovative thing you’ve ever done?

Every significant innovation that I have accomplished in my career has come from applying concepts from one field in a new and unique way to the problem at hand. This core technique can be traced back to a formative innovation I made as a graduate student studying oceanography. During my very first research expedition I found myself huddled up in a small radio shack at 4 in the morning listening to underwater sonar pings straight out of a submarine action movie. Ping….(pong). The radio crackled, and the third mate asked, “Are we ready to move on to the next waypoint?” I picked up the mic and responded, “Yes please. Two more to go.” We were 18 hours into our expedition and I was the only scientist awake, wondering how on Earth a 23-year-old grad student was trusted to take charge of a research ship in the middle of the Pacific. I sure hoped the modifications I made to the ray tracing algorithm we had adapted from a different field did its job!

Oceanographers use all kinds of embedded sensors and technology to measure the ocean, which is always moving, changing, and keeping you on your toes. Knowing your location in time and space is critical for every type of oceanographic expedition, and in this case I had been put in charge of improving the underwater navigation algorithm for the submersible Alvin. Beyond a basic exploration and sampling expedition, we planned to use Alvin to create a high-resolution map of the seafloor. Because radio waves cannot penetrate water very far, underwater vehicles use acoustic navigation to locate themselves instead of GPS. During my first year of graduate school I used a new ray tracing algorithm to improve Alvin’s navigation. It worked quite well, reducing navigation drop-out considerably and providing an accuracy of 2-3 meters. Since Alvin’s lights only penetrate about 20 meters into the darkness, this improvement was critical to improved mapping and efficiency during the expedition. The specific algorithm I ended up using wasn’t the most important thing – the application of that algorithm and processing chain in a new domain was the real innovation. Recognizing how to apply techniques from one field in a new way in a different field turns out to be one of the most useful things I have learned so far in my career, and I have often been able to use that technique when faced with a tricky problem to solve or customer issue with no obvious solution.

After graduate school I spent several years teaching, starting and running a non-profit and doing research as a scientific programmer at the University of Washington. My specific area of research was connecting different numerical modeling systems to live data feeds, and this experience got me recruited into a renewable energy information startup called 3TIER. Initially hired to be an analyst and research programmer, I was quickly repurposed to work on a web interface for generating automated reports about renewable energy potential given a certain location.

Instead of using a form to determine the location of interest, we created a mashup of Google Maps and a color overlay of wind potential. Once again, I was able to apply an algorithm from a different domain to create image tiles that laid over the Google map in a performant manner and allowed for a simple yet powerful visualization. This visualization also simplified the user experience by allowing the user to click directly on the map instead of entering the latitude and longitude by hand, which was error prone.  We launched the interface 6 months later and as we went to market I moved into the Product Manager role. I quickly realized the role was a good fit for my technical and teaching skills. Our product, FirstLook, went on to win an industry award from the American Wind Energy Association (AWEA Commercial Achievement Award, 2009). 3TIER was later acquired by large weather instrumentation company, and the core bits that I developed are still in use today.

Hoping to broaden my product management experience, I joined a large cooperative of libraries that builds different types of library technology called OCLC. I was hired to work on an interesting product called CONTENTdm, a content management solution for libraries, archives and museums. When I became the product manager for CONTENTdm I learned the development team was burned out from a long overdue product release that had gone poorly. We adopted scrum and agile development and I worked as the product owner. I was confident that adopting agile processes would solve many of the issues the team was encountering, both on the software side as well as the product side. I got out of the building, built quality use cases and helped empower the team to improve. This time my innovation was around process, but shared the common theme of applying a known solution from somewhere else in a new and useful way.

The overall transformation for the product and the team was significant. After my first year with the team, we went from a 6-9 month release cycle to full releases every quarter and publicly visible updates every 2 weeks on our staging server. Even though OCLC considered the product to be in “maintenance” phase when I started, our improved execution and feature enhancements led to an increase in customers of 25%, with the majority of that growth in our high margin cloud based offering, raising our installed base to over 2500 installations of CONTENTdm.

In every case where I am challenged with a new and interesting problem, my mind returns to that late night of pings and pongs and the satisfaction I felt diving in Alvin with a greatly improved navigation system. Is there a different way to do this? How can I solve a customer pain point they don’t explicitly realize they have? What ground work needs to be in place for such a solution to work correctly? How can we simplify that solution? What surprises are we likely to encounter? Between my scientific training and teaching DNA I can’t help but to ask questions, stay curious and solve problems.