Duolingo's language learning app employs a competitive leaderboard system, where every week learners have the opportunity to advance through 10 different leagues. While some learners invested lots of time trying to advance to the top leagues, others tended to stagnate in the lower leagues and engage less with the app overall.

Throughout the summer, I was tasked with developing a product roadmap to increase lower league learner engagement. I conducted UXR with 200+ users through a combination of a survey and live interviews to extract motivational aspects and pain points of the leaderboards feature. I also analyzed several experiments and key growth metrics in context of users in lower leaderboard leagues to determine how much impact addressing these learners could have. By the end of the summer, I had ideated 20+ new leaderboard features targeting aforementioned learners, collaborating with engineering and design to determine the investment necessary to develop a rudimentary prioritization scheme by ROI. I launched one of these features as an experiment, resulting in +20,000 daily active users, and put another one in the development pipeline. I presented my research findings and roadmap highlights to Chief Product Officer and Growth, Monetization, and Learning area leads.

Especially during and after the pandemic, remote work and collaboration increased significantly. As a result, more organizations were inviting guest employees from external sources to utilize their resources for collaboration. However, guest users - especially those that are stale or over-permissioned - can serve as potential security vectors into a host organization.

Throughout the summer, I worked on a project to address this problem. I ideated 10+ analytics to surface information about this problem and tested them on customer data. Concurrently, I demoed in-progress analytics capabilities to 5 customers, utilizing their feedback to iterate and create new analytics. I worked with a UX designer to mock up a prototype of the product and translate it to product-quality wireframes, and managed a software engineer and intern to develop two scaled-down analytics for initial testing. I also collaborated with 5 other program managers to set the stage for incorporating different aspects of previous products with my summer project.

At Vianai, I performed competitive analyses of 60+ AI technology companies in terms of feature offerings & industry verticals, choosing an additional 10+ for user experience focus. I also researched multi-criteria decision analysis, including its application areas and solution categories, types and methods, developing example demonstrations as well. On the technical side, I also cleaned and compiled phone and email communication data for a client company for the data scientists on my team to perform advanced inference. I presented all of my work throughout the internship to the Head of Platform and several principal data scientists and product managers.

Clubfeast is a subscription-based restaurant delivery app in which customers subscribe weekly to a predetermined number of deliveries for a predetermined number of people, order meals at least one day in advance, and receive meals in a predetermined delivery window. Clubfeast delivers restaurant meals 40% cheaper than competitors and addresses inefficiency in restaurant kitchen and ingredient usage, food delivery, and customer expenses.

As an associate product manager, my role entailed optimizing and improving elements of the customer interface, directing coders, and testing changes. As lead UI/UX designer, I used Adobe XD to wireframe and prototype the customer mobile app, and directed early-stage code development.

I decorated 100+ cupcakes and 50+ full size cakes every week to order.

I worked under Dr. Martin Elvis to determine the potential heights and power generation capabilities of solar power towers built on the Moon. This project built off of an earlier paper by Dr. Elvis, which discusses the lunar south pole's Peaks of Eternal Light (very small areas with virtually constant solar illumination and thus solar power) and a valuable resource (water) in a permanently dark crater. However, because building high increases the amount of continuous illuminated area, building tall towers on the lunar south pole is of interest for resource sharing and cooperation between countries in the near future of space exploration. Thus, we aimed to calculate within reasonable assumptions how high a solar power generation tower could be built on the lunar south pole, as well as how much power such towers could generate.

My role specifically entailed utilizing Python to map solar illumination of lunar south pole at various heights above the surface, graph illuminated area at various illumination intervals, analyze trends of solar illumination through monthly oscillating angles of illumination, determine estimated power generation capabilities from the towers. A first-authored paper titled "Quantifying the Available Solar Power near the Lunar South Pole" is in preparation.

My team at the LIT lab created a multimodal learning analysis website for a Fall 2020 graduate course to replace the usual modal sensors students would receive during in-person learning. Specifically, I used HTML, CSS, and Javascript to integrate an eye-tracking modality into the website and standardize the modality to conform with the other tools.

I was on the teaching staff for Introduction to Computer Science with Python. I taught twice-weekly sections to 10-20 students regarding computer programming fundamentals, Python basics, and problem-solving techniques. In addition to sections, I hosted twice-weekly office hours to further explain and walk students through specific conceptual and homework-related questions to students. I also graded hundreds of tests and problem set submissions on a weekly basis.

Working directly under the Chief Strategy Officer at Tempus, my internship team created a database of 260+ oncology-focused biotech company profiles, including financial information (IPO date, market cap, funding rounds, valuation, etc.), personnel (founders, executives, board of directors), and assets & pipeline (mechanism of action, target indication, subtypes/genes of emphasis, clinical trials). We researched 18 healthcare-focused VC firms to further build our list of companies. We also analyzed and visualized trends in the mechanisms of action, target indications, subtypes/genes of emphasis, and novelty of the 800+ programs we profiled. We presented our findings to the Chief Strategy Officer, Chief Science Officer, Chief Medical Officer, Senior Vice President of Data Science, and Director of Operations.

I worked under the guidance of a graduate student, Katie Breivik, and Professor Vicky Kalogera. My task was to create a database of neutron star and black hole binaries from different binary evolution models; the database was to be used to investigate how different models influence the number and properties of these binaries observable by LIGO. Under their instruction, I used a computer program called COSMIC (Compact Object Synthesis and Monte-Carlo Investigation Code), which in turn utilized BSE (Binary Star Evolution code), to simulate binary star evolutions and record the occurrences of three outcomes (neutron star-neutron star binaries, neutron star-black hole binaries, and black hole-black hole binaries).