CS Night - 2020
Join the Department of Computer Science for the 19th Annual Computer Science Night
Thursday, Dec. 3 | 6:00–8:30 p.m.
Register now
Program Schedule
5:30pm - 6:00pm - Setup Hopin and join the event
6:00pm - 7:00pm - Student project presentations
7:00pm - 7:30pm - Awards Distribution & Address by the Department Chair
7:30pm onwards - Networking
This year's research and project abstracts
Graduate Projects
SnapLogic UI Generator
Presenters: Rubin Johny, Suryadeep Guha & Tushar Sehgal
Sponsors: Greg Benson, Cole Howard & Wenhan Rao, SnapLogic
Faculty Advisor: Prof. Olga Karpenko, CS690
SnapLogic is an Integration Platform as a Service (iPaaS) that enables quick development of sophisticated data and application pipelines using a visual drag and drop interface. The resulting pipelines can be scheduled to execute on a regular basis or they can be exposed as HTTP REST APIs (serverless microservices) with sophisticated policies through API management (APIM). We have developed a new feature within SnapLogic that allows pipelines to become interactive web user interfaces. These UIs are powered by the pipeline itself and the SnapLogic APIM framework. The application can dynamically render UI components making use of custom conventions defined within the pipeline. This feature enables a non-technical business user to interact with SnapLogic pipelines through a web application deployed using a single click on the SnapLogic Platform.
SnapLogic Persistent Event Streaming
Presenters: Nilay Narlawar, Vaishnav Kshirsagar, Ameya Yerpude
Sponsor: Professor Greg Benson, SnapLogic
Faculty Advisor: Professor Olga Karpenko, CS690
SnapLogic is an Integration Platform as a Service (iPaaS) that enables quick development of sophisticated data and application pipelines using a visual drag and drop user interface. The current SnapLogic execution engine does not provide durable and persistent storage for in-flight data, instead, most processing is done in memory. We developed a prototype of a SnapLogic Persistent Event Streaming service that extends the existing architecture by providing a native streaming service. Ultimately, this service can be integrated with the native Snaplex execution environment so that SnapLogic customers can easily deploy and leverage event streaming. In this project we built Kafka-like components, a durable and persistent logging mechanism, and distributed consensus. Our implementation can survive node failures. We also build Streaming Service Snaps that allows a SnapLogic user to build visual pipelines that can read and write to our streaming service.
Nonprofit Donation Platform
Presenters: Roger Yu, Na Lu, Danielle Hu
Sponsor: Jose Alvarado, SF Dev Shop
Faculty Advisor: Professor Olga Karpenko, CS690
Nonprofit Donation Platform (NDP) aims to help democratize philanthropy through individual giving. It makes giving to your favorite nonprofit organization safer, faster, and easier than ever. NDP is open to licensed nonprofit organizations and individual users passionate about philanthropy. Our platform is available on both desktop and mobile devices, so users can donate from anywhere. Every user will be able to see a history of all their donations as well as have a public-facing profile similar to LinkedIn displaying all their friends’ donations as well as those of the public. Our platform is also unique in the way that donors can view the status updates of all the organizations they follow.
With our easy-to-use platform, nonprofit organizations can leverage modern marketing strategies to find new donors and do the most good. Donors can also leave a note along with their donation, expressing themselves towards the nonprofit organization, so other in-network donors can join them and make the society better.
PNC Bank Interest Rate Risk Platform
Presenters: Tracy Cheng, Yan Liang, Yuyao Xie, Siyu Zhao
Sponsors: Christopher Whittelsey and Michael McLaughlin - PNC Bank
Faculty Advisor: Professor Olga Karpenko, CS690
Risk Managers analyze complex financial data with the help of data scientists. One of the many requests of the data scientists is to explore the interest rate sensitivity (IRR) of the firm - a process that takes hours, sometimes days. This back and forth communication causes inefficiency and delay in the management IRR. This delay in response is further exacerbated by current working limitations caused by COVID-19 safety concerns and motivates the deployment of tailor-made productivity tools.
To solve this problem we built Gamma10, a web application that allows financial managers and advisors to easily create, execute, and visualize interest rate risk scenarios in a matter of minutes. We believe our product will help serve as a bridge between finance and technology, making financial analysis more accessible for non-data scientists.
Company Disambiguation Tool
Presenters: Anna Jurgensen, Divya Vijayan, Zirong Wang
Sponsor: David Buttler, Lawrence Livermore National Lab
Faculty Advisor: Professor Olga Karpenko, CS690
Researchers and journalists trying to thoroughly study and analyze companies have several challenges to contend with. First, a company can be referred to with many different name variants, such as formal names used in litigation or official documents that differ from shorter names used in marketing, internal documents, or in reference to that company in news articles. Second, companies in different industries may have similar names even if they aren't actually related, and it can be difficult for a researcher to determine whether or not they should consider a name as a variant for the company they are studying without first manually searching for information about that name itself. As such, these researchers face a time consuming and tedious process of identifying all possible name variants for a specific company, and eliminating false leads. In this project, the team created a web application that uses NLP to help researchers more quickly and easily work through their process of company name identification and disambiguation.
Pocket Realtor
Presenters: Junting Cai, Jerry Chen, John Fox
Sponsor: Jose Alvarado
Faculty Advisor: Professor Olga Karpenko, CS690
Pocket Realtor is a platform for property owners, real estate agents, and tenants. It seeks to make the process of renting easier than ever before. Property owners can add their properties to the platform. Real estate agents can contact property owners who are interested in selling their properties. Tenants will discover useful information about available properties to rent in their area, such as square footage, rent price, nearby amenities, and more. Not only that, tenants are able to create a group with their friends so that they can share information about their rental choices with one another and find a place that is the best fit. Tenants can schedule in-person showings and contact property owners for more information about their rentals. Pocket Realtor is your all-in-one solution to finding a great place to live, rent, and sell.
Energy Aware Datacenter
Presenters: Siddhartha Chowdhuri, Sum Xu, Tong Wang, Lacee Xu
Sponsor: Malini Bhandaru & Pranay Pareek, VMware
Faculty Advisor: Professor Doug Halperin, CS690
As demand for Data Center services has grown exponentially so too has the energy consumption of the underlying infrastructure. Increasing energy costs and environmental considerations have combined to elevate the importance of reducing data center energy consumption. In our project, we address this by architecting an energy-aware Data Center based on the logic that not all jobs are equally important or need to be run instantly in parallel in a Data Center. We built a controller to manage workloads within an energy budget by consolidating workloads on fewer machines, dropping non-essential jobs, dropping duplicates, and migrating workloads across Data Centers. We use Kubernetes, a technology that is lighter weight than a virtual machine while also providing greater bin packing even as it shrinks an application to ensure successful deployments. We also use Golang, Terraform, and React in building a Kubernetes energy-aware datacenter in GCP.
Sweeply
Presenters: Aishwarya Parthasarathy, Jaya Lakshmi, Radha Prashanthi, Aakula Chinna
Sponsor: Dustin Lyons
Faculty Advisor: Professor Doug Halperin, CS690
Sweeply -- a Slackbot helps organize very large channel lists within an organization’s Slack environment. After using Slack for several years, a user’s channel list may grow unwieldy and become difficult to effectively utilize. Slack does not provide tools to actually leave channels that are no longer of interest; Slack only allows a user to hide them. However, for someone who is a big proponent of "inbox zero" it’s important for them to manage channels in an effective manner. Sweeply is designed to put the user’s channel list on "autopilot". By providing and implementing a set of rules and criteria for how users joins and leaves channels, Sweeply automates the process for such users.
Estimating Claim Likelihood on Extend Warranty Contracts
Presenters: Daniel Havir, Tyler Iams, Zini Zhu
Sponsor: Michael Darmousseh & Mohnish Mohanty, Extend
Faculty Advisor: Professor Doug Halperin, CS690
Extend offers extended warranties as a service for eCommerce platforms through an API endpoint. In this project, we aim to help Extend’s pricing team more effectively assess the risk that a service claim would be filed for an extended warranty. We built a simple and interpretable probabilistic approach to estimate the likelihood of experiencing a claim on a warranty using historic data together with available open data. On top of this approach, our team built a light-weight web application to help visualize the data as well as understand the model’s performance in the pricing process.
A Multi-Destination Route Optimizer for Travel Planning
Presenters: Aarthi Parthipan, Sukirti Dash, Madhuri Devi
Sponsor: John Huang, One Hundred Feet Inc.
Faculty Advisor: Professor Doug Halperin, CS690
Our team has developed a personalized travel planning app that generates optimized multi- destination routes – walking and driving – based on user preferences for places they want to visit in a city, travel dates, time they want to spend at each place, and user place ratings. This capability is not available in any of the travel planning apps in the market and makes travel planning easier and more personalized. We solved the route optimization problem, addressing complexities like time constraints, multiple modes of travel and user preferences. This is a mobile-first Android application and is driven by a microservices based architecture that leverages integration with Google APIs.
Middleware Microservice for Openprise
Presenters: Aswin Pottenplackal, Hrishikesh Mulabagula, Jie Liu, Xinqi Zhang
Sponsor: Mario Lim, Openprise
Faculty Advisor: Professor Doug Halperin, CS690
Exposing data from a database via APIs can be quite tedious and time-consuming. The goal of this project is to facilitate automated creation of such an API whenever users look to access their data through a REST or GraphQL API. We build a system that allowed data administrators to set data access policies via a web UI and generate corresponding data access API (REST and GraphQL ) automatically. At the same time, the system auto-generates documentation that describes how to use the newly created APIs.
Myrror - Smart Home Product
Presenters: Eric Lieu, Nishant Mehta
Sponsor: Nishant Mehta
Faculty Advisor: Professor Doug Halperin, CS690
Myrror is the future of IoT products – a smart mirror that's a beautiful piece of furniture. It provides a digital dashboard that can show you the time, your calendar, YouTube, and act as a central point for IoT products like Ring video camera, Alexa, Google Home, Nest Temperature, and more. What once existed as a niche DIY engineering product is now accessible to everyone. To personalize and interact with your mirror, we've created an intuitive React-Native mobile app that allows the non-technical user to manage everything from the face of the mirror – positioning items, installing new modules, and pushing remote updates with new features.
Factbook
Presenters: Ananya Balakrishna Shetty, Ayush Arora, Mitali Sawant
Sponsor: Jon Rahoi
Faculty Advisor: Professor Doug Halperin, CS690
FactBook is a user-friendly web application to enhance users' knowledge about world. The website's main objective is to display the available data about countries in an organized and accessible manner to increase user engagement and simplify exploration. The website provides a faceted search section (similar to that found on eCommerce websites) allowing users to filter and identify countries based on specific categories such as all the countries in the continent of Africa or all the countries that use the Euro. The website displays the identified countries in real time on a world map. FactBook also includes a quiz section to increase user engagement.
Automated Modeling Platform
Presenters: Jiaye Qin, Jiachen Liu, Meng Lyu
Sponsor: Christopher Whittelsey and Michael McLaughlin, PNC Bank
Faculty Advisor: Professor Olga Karpenko, CS690
We worked with our project sponsor to enhance their ‘Automated Modeling Platform’ (AMP). The platform removes the necessity of modelers to be programming experts by allowing them, via a web application, to execute various models, store data and compare the results. The web application removes the risk of coding errors and mathematical errors. Our primary enhancements were to introduce detailed descriptive modeling statistics, allow for a more robust variable selection process and include a way for the user to easily compare various models. We also improved the aesthetics of the UI and provided actionable recommendations on improving the code efficiency. The resulting product is a production ready modeling platform that will support the bank’s financial modeling efforts.
A Live Information Analysis System Powered by Deep Learning
Presenters: Seung Jung Jin, Sami Ullah, S.Masoud Hosseini
Sponsor: Weidong Yang, Kineviz, Inc.
Faculty Advisor: Professor Olga Karpenko
Information is not an isolated piece of information; instead, information is rich with connections to different pieces of information. By exploring the connections between different parts of the data we can find out interesting stories hidden between facts which might contradict with what we perceived to be true. Through our big data workflow that connects with Deep Learning (DL) and (AI) technologies, we are able to greatly reduce the difficulty of collecting, augmenting, and analyzing data at scale. We’ve adapted this project on the Presidential Election 2020 and have come across various analyses that forecasted a very competitive election.
Ethereum Data Analytics
Presenter: Simon Lu
Team Member: Jinyue Song
Faculty Advisor: Professor Eunjin (EJ) Jung
Ethereum has recently regained its popularity due to the emergence of DeFi and the growing anticipation surrounding Ethereum 2.0. It is crucial to understand the reliability and usability of Ethereum given its expanding user base and market capitalization. This project analyzes the patterns of blocks and transactions in the Ethereum network using newly collected from 2019 to 2020 and compares the analysis with findings from previous work. Through these analyses, we discovered that Ethereum has become more usable as fewer transactions are abandoned. However, the overall number of forked/orphaned blocks has decreased, which suggests an increase in centralization among the miners.
Undergraduate Projects
Rapid Website Creation Studio
Presenters: Priscilla Hamiter, Ahmaad Idrees, Alejandro Marcovich
Sponsor: Roger Ruttimann, Reactful
Faculty Advisor: Professor Doug Halperin, CS490
Rapid Website Creation Studio is an interactive React-based web application for designers and marketeers to create fully customized content for each site visitor based on the visitor’s unique characteristics and on-site behavior. The implementation of dynamic components allows for a more personalized user experience, resulting in websites with greater levels of engagement. This project adds to the RWC Studio application the capability for non-technical users to create and maintain HTML fragments which can then be dynamically rendered to assemble a webpage, based on content labels, that result in higher levels of user engagement.
Animal Colony Management
Presenters: Cyrus Robins, Isabel Ringo, Orshi Kozek
Sponsor: Alark Joshi, USF Computer Science
Faculty Advisor: Professor Doug Halperin, CS490
Researchers doing animal studies typically use spreadsheets to keep track of their animal colonies, a process which is both laborious and error prone. To address these issues, we have continued work from prior semesters on an easy-to-use React-based web application that allows researchers to keep track of their colonies online. Based on input from researchers managing animal colonies, we have integrated a variety of features that support research and cut down on costly mistakes. These include rigorous error-checking to ensure that incorrect data is never inputted into the system, a comprehensive search algorithm that makes it easy to navigate and manage large animal colonies, and an improved image gallery that allows users to easily sort and manage photos of their colonies. We believe that with a little more fine-tuning, this application could be used in the near future by scientists around the world.
AccesSOS Progressive Web App Enhancements & Text-to-911 Map
Presenters: Alexandria Davis, Annika Rodriguez, Denali Webber, Maria Paula Linares
Sponsor: Gabriella Wong, AccesSOS
Faculty Advisor: Professor Doug Halperin, CS490
The primary goal of AccesSOS is to provide access to 911 emergency services via a React-based Progressive Web App (PWA) for those who can’t hear or can’t speak out loud or are non-English speakers. The USF team made the PWA more accessible for users and first-responders by slowing down text-to-voice speech, creating a visual friendly location pin on a displayed map, and directing users step by step through the app’s functionality. In addition, the team created a page – already in use at http://text911.info – that displays up-to-date information on the availability of text-to-911 services across the country.
Communicating with Memes
Presenters: Jake Ols, Bradley Genao, Alex Kowalczuk, Ned Azar
Sponsor: Jose Alvarado
Faculty Advisor: Professor Doug Halperin, CS490
Memeapp is a social network application designed to bring people together through sharing of memes. Many want to explore memes, but it's hard to find a practical and easy-to-use app just for memes. We designed the app to be as simple as possible with a user-friendly interface that is easy to learn. Our application allows users to swipe memes in a Tinder-like fashion and share those memes that they think are funny. There is currently no other solution that caters only to memes, or any other image-sharing app that leverages the use of swiping gestures. The front-end is built entirely using React Native and will work for both Android and IOS. The backend leverages JavaScript and PostgreSQL.
BART - Platform Signage
Presenters: Ahmed Kaddoura, Antonio Gutierrez, Summer Marsh
Sponsor: Herb Diamant, BART
Faculty Advisor: Professor Doug Halperin, CS490
Our goal is to support BART engineers by helping to build a system to identify and correct platform sign inaccuracies. Currently, there is no system that actively monitors the accuracy of platform signs, so engineers are blind to problems until a customer files a complaint. We took a data-driven approach to improving platform sign accuracy by using Splunk to process and analyze BART train log data to identify recurring conditions associated with sign inaccuracy. We also used Splunk to build dynamic dashboards and interactive data visualizations as tools for BART engineers to examine these data. We hope our project will help improve the experience of hundreds of thousands of Bay Area residents who use BART.
Healthcare Dashboard
Presenters: Jochem Dierx, Aditya Dixit, Analucía Esparza-Chávez
Sponsor: Tim Castillo
Faculty Advisor: Professor Doug Halperin, CS490
We used React, D3, and ChartJS to build a Healthcare Dashboard website to visualize data from the MIMIC-III Clinical Database in a simple and interactive way. The Dashboard’s visualization capabilities provide easy access to clinical data for healthcare researchers, hospital administrators, and data enthusiasts who do not have a computer science background, for example individuals seeking to obtain insights from correlations between different clinical variables or through other data-based analyses that can be visualized.
UI Selectors Updated
Presenters: Hanglin Li, Serena Pang, Stew Powell
Sponsor: Jeff Johnson, USF Computer Science
Faculty Advisor: Professor Doug Halperin, CS490
UI Selectors is a user-interface library for developers to build applications that make use of semantic-based controls. The focus in UI Selectors is not on how the control in a user interface looks (e.g., the appearance of a button), but rather on its underlying function. It is intended to eliminate the possibility of a programmer choosing a presenter that does not correctly represent the desired function. Our team updated UI Selectors to Java Standard Widget Toolkit(SWT) and built One-from-N, Many-from-N, On/Off, and Value-in-Range controls. UI Selectors is a unique approach compared to other UI toolkits like Figma, Sketch, or Adobe XD in that it serves as a platform of semantic UI controllers at the presentation layer for those toolkits.
Expandi
Presenters: Di Zhang, Jason Wahjudi, Stephen Yu
Sponsor: Jon Rahoi
Faculty Advisor: Professor Doug Halperin, CS490
Expandi is a web version of the popular board game, Gaia Project. In Gaia Project, 2 to 4 players earn Victory Points by building and upgrading various structures, performing research, forming federations, and terraforming other planets. Expandi was previously developed by a CS690 group and included features such as multiplayer and game history. In this newest version of Expandi, we have implemented a mechanism to authenticate users in online games and to save and load games, and we have also enhanced various features of the game to make it more fun and intuitive for everyone.