Building a Smart Grid Interface Before Everyone Was Doing It
One of the core decisions power companies need to make has to deal with power generation and grid switching. The companies need to provide power to their clients; they can either buy it from the free market or generate it themselves. Most power generation comes with its own costs (like buying coal for a coal plant), or has weird conditionals (like wind generation depending on the weather), or has scheduling requirements (like natural gas plants take a long time to warm up, so you never turn one on unless it's going to be running non-stop for at least a few weeks). The goal is to look at the estimated load on the system, the price of electricity on the market, and all your personal system requirements, then to figure out the cheapest possible combination of things you can do.
A few years ago, our client was working on a new system using agent-based modeling and machine learning to solve this problem. Their system was producing better (ie cheaper) results than the previous methods. They partnered with UpThought to design and develop a slick web interface that would let them see the current state of the system in real-time, historical states of the system, and provide insight into how their modeling was performing.
Visualizing Emerging Trends in Geographic and Political Data
A state-owned enterprise (SOE) is basically a legal entity that governments use to conduct business. Our client is maintaining data on all of the SOEs in some of the world's economic super powers. They track office locations, investments, transactions, currency exchanges, bargains, etc. This is a lot of data.
UpThought helped them figure out the most useful ways to display the data, implement searching and filtering, and generate analyses and/or heuristics based on current and historical trends. We built a system that takes critical, valuable data, and visualizes it in meaningful and insightful ways. Specifically, we designed and implemented multiple visualizations of the data, and built analytic and heuristic models that tie into those visualizations to help expose emerging trends and relationships.
Crowd-sourcing the Transcription of Some of America's Most Treasured Documents
The National Archives wanted to crowd-source the transcription of some of America's most important historical documents. The project featured over 300 documents (about 1,000 pages) ranging from the late 18th century through the 20th century, including letters to a civil war spy, various acts and laws, presidential records, suffrage petitions, indictments, and fugitive slave case files.
UpThought designed and developed an interface to allow citizen archivists to browse documents by difficulty (beginner, intermediate, or advanced), review the documents, and collaborate in the transcription process.
The project was nominated for a Webby, ultimately losing out to the overachievers at NASA.