Cobblestone Energy is an energy trading company that provides keys insights on European power supply and demand, empowering investors in the U.K, Netherlands, France and Belgium markets to buy and sell power. Our client needed a big-data pipeline solution to deliver and cache reliable, real-time and actionable results at minimum latency with an intake of over 1 million records per minute.We built this bespoke data processing pipeline on an AWS infrastructure. We kicked off the project with data collection by Lambda function and then stored, cleaned, and transformed the data to create a new cache of raw data. Next, we populated this new data to the Grafana visualization tool, which converts the data into user-friendly infographics. Finally, we implemented a data validation step to ensure the accuracy of our results.

Global chemical corporation With over 20,000 employees at sites around the world. Myriad conditions can impact the efficacy of commercial chemical reactions at scale, from minute changes in the manufacturing environment to the composition of the various gasses that are combined at different points in a process. Our client needed an accurate tool to measure and report on real-time conditions and optimization ratios throughout their manufacturing systems, instead of relying on the experience, intuition and manual reflexes of factory operators. THE PIVOTAL SOLUTIONFirst, our team designed and built a predictive ML software model and tested our solution online; the result, a new automated gas-balancing optimizer feature, was then embedded into our client’s chemical manufacturing processes. Our solution takes the guesswork out of gas balance optimization by recalculating the optimum balance once every 60 seconds for all pipe flows in the system, allowing operators to make precise adjustments as needed.

New Forest Film uses extensive market research and an agile development model to produce dynamic and deeply targeted media content. Our client wanted to leverage their ongoing audience survey program to produce media content with a built-in audienceBased on the data provided by the client, we built several different clustering ML models. From these models, we selected the model build that produced the best result in terms of clusterization quality and ROI. We also delivered cluster visualizations and results analysis.