Aeternum’s mission is to make air quality more equitable. We do this with our low-cost, high accuracy, communicating air quality sensor networks and cloud-based data analytics. Enabling wide area deployments, Aeternum solutions provide objective insights and transparency around the effects of pollution on human health and on the environment where people live and breathe. We help stakeholders — civilians, municipalities, businesses — make informed decisions about air quality. My role at Aeternum was to deliver data reports to partners/customers and to lead the calibration effort of the sensor data. Through this experience, I am intimately aware of the fact that data on its own has little value. It is necessary to uncover and extract patterns, trends, and anomalies in the data and deliver it to those in the positions to affect change. Every project is different and each environment in which the sensors are deployed varies. The pollution sources vary, the meteorological conditions change, and therefore the pollution mitigation strategies must be tailored to each project. I analyzed and visualized the data collected by Aeternum air quality sensors and delivered these insights in the form of reports. Communicating data in an effective and actionable way is a critical step in the path towards cleaner air. Another responsibility that I undertook was calibrating the sensors such that the data is accurate and reliable. Collecting data that stakeholders can trust is absolutely critical to Aeternum’s mission of making air quality monitoring ubiquitous, accessible, and low-cost. As a master's in statistics student at Columbia, I have developed many technical and soft skills that I was able to apply to my work at Aeternum. I directly applied learnings from courses such as Machine Learning, Statistical Computing, Data Visualization, and Linear Regression to my tasks. Machine Learning, in particular, is a field that will transform the power of these low-cost air quality sensors. Aeternum’s low-cost, easily deployable sensors will accelerate the amount of air quality data collected, thereby expanding the research potential of the space and necessitating the use of machine learning. I was able to employ ML techniques to improve the accuracy of the sensors and to extract patterns from the data. Moreover, I utilized the learnings from statistical computing and statistical graphics courses to deliver insightful and effective project reports. Overall my degree has provided me with the analytical and statistical foundation to build on and apply to a new domain. The air quality sensor space is still very much in its infancy. The traditional approach to monitoring air quality involves installing and managing large, expensive stations that are well-understood but do not provide enough data resolution in space or time. On the other hand, Aeternum’s air quality sensors rely on new technology that’s continuously being improved and still has enormous potential. As a result, often we are charting new territory, which can be both exciting and daunting. When the path isn’t already paved, success is undefined and it’s not always clear whether your next step is even in the right direction. Often times, the sensor data illicits more questions than answers — you develop hypotheses and ideas, but there is no one-size-fits-all approach or solution when it comes to air pollution. The research side of the air pollution space is growing, but we are still only scratching the surface in terms of our understanding. This makes the work at Aeternum all the more valuable and important. Through this experience — working in the business world, at a startup, as a data scientist — I gleaned many personally valuable takeaways. To start, I have truly appreciated that business, no matter how process-oriented or data-driven it is, is fundamentally composed of and advanced by people. People are the ones with the ability to affect change, and they define the success of a product, business, solution and society. Related to this, my perspective of data has been enhanced through this experience. Oftentimes, when we think of data, we think of cold, hard objective information. However, the data we see on the screen is impacted by a series of (human) decisions. Humans define how the data is collected (which often results in biases), how often it is collected, how to define certain variables, how to correct the data, etc. All of these decisions impact the information in front of us and the conclusions we draw from it. In line with this idea, another pivotal conclusion I’ve drawn is that data on its own has little value, instead communicating the information is where value is formed. In the case of air quality data, it is most impactful and actionable when insights are conveyed clearly to the stakeholders with the power to have a positive impact.