Clinic Projects

Please click on a link below to view the Harvey Mudd College Computer Science Clinic projects for the corresponding time period.

Clinic Projects for 2015-2016

Innovative Graphics and Input Enhancements for the Satellite Orbit Analysis Program (SOAP)

Client
Aerospace Corporation

Faculty Advisor
Geoff Kuenning

Student Team
Eric Caldwell, Zakkai Davidson, Christian Guerrero, Fabiha Hannan, and Emily Stansbury

Joint project with the engineering department

The Aerospace Corporation is a Federal Funded Research and development Center (FFRDC) that provides its Satellite Orbit Analysis Program (SOAP) to a variety of national security and space customers. This clinic project encapsulated a number of 3D rendering techniques in a digital globe framework that can be readily adapted for incorporation into SOAP and/or other OpenGL-based graphics software. The supported effects include real time weather data acquisition and visualization, terrain shadowing, smoke simulation, improved trajectory drawing, and realistic night and day transition.

Predicting Customer Behavior from Credit Card Transactions

Client
American Express

Faculty Advisor
Yi-Chieh (Jessica) Wu

Student Team
November Baez, Kelly Lee, Maury Quijada, Sarah Trisorus, and Paula Yuan

American Express is a global financial services company with over a trillion dollars in card member spend. In order to improve the experience of American Express card members and merchants, the Clinic project uses this spend data to predict future customer transactions at specific merchants. To achieve this, the project documents and designs a framework using various machine-learning models that yield accurate transactional predictions for American Express card members and merchants.

Indoor Mapping with Bluetooth® Beacons

Client
Gimbal

Faculty Advisor
Melissa O’Neill

Student Team
Kaitlyn Anderson, Louie Brann, Paige Garratt, and Cyrus Huang

Gimbal is the producer of the most widely deployed Bluetooth® Smart beacons in the world. With this technology, Gimbal provides a solution that allows users to receive relevant location-based content, directly to their mobile phones. However the beacons need to be placed properly in order for them to be effective. We developed an iOS application that uses the phone's sensors as well as signals from the beacons to create an accurate, interactive map of the installed constellation of beacons.

Location-Aware Date Sharing Between Devices

Client
Intel Corporation

Faculty Advisor
Robert Keller

Student Team
Anne Christy, Daniel Cogan, Hugo Ho, and Emma Meersman

We designed an approach to using location-based data and querying as a means of exploring new ways for user and devices to interact over the Internet. We created an application programming interface (API) that provides convenient tools for developers to integrate location-based communication in a wide range of applications. We developed applications on Android phones that use our API to demonstrate and validate our approach.

Simple Real-Time Data Analysis System

Client
Laserfiche

Faculty Advisor
Beth Trushkowsky

Student Team
Rachelle Holmgren, Amit Maor, Drew Schmitt, and Jean Sung

Laserfiche provides its customers with a system to create, execute, and report on business processes. The goal of our project was to create a data analysis system that provides each customer with insights they could use to improve their processes. Our team researched and tested various methods for analyzing business process data that would be beneficial to the wide range of Laserfiche customers.

Understanding Detailed Memory Performance Data

Client
Lawrence Livermore National Laboratories

Faculty Advisor
Chris Stone

Student Team
Paul Dapolito, Andrew Fishberg, Samuel Jackson, and Xiaotian Wang

Mitos and MemAxes, a pair of memory analysis and visualization tools developed by Lawrence Livermore National Laboratory's Scalability Team, aid supercomputer users in diagnosing high performance computing (HPC) speed bottlenecks. The LLNL Memory Clinic Team has expanded these tools to be more accessible and powerful for users. Specifically, the team has augmented widely used physics mesh simulation programs to provide detailed Mitos traces and explored machine learning techniques for automatically diagnosing performance problems.

Atomistic Simulations of White Dwarf Dynamic

Client
Lawrence Livermore National Laboratories

Faculty Advisor
Chris Stone

Student Team
Phillip Diffley, Daniel Houck, Lennart Rudolph, and Skyler Williams

Joint project with the physics department

Lawrence Livermore National Laboratory (LLNL) wishes to better understand the liquid- crystal interface found in white dwarfs. Using LLNL's massively parallel molecular dynamics simulation code (ddcMD) along with the Blue Gene Q supercomputer, the HMC team will simulate a small region of the interface to shed light on impurity sedimentation and mixing.

Digital Aging

Client
MITRE Corporation

Faculty Advisor
Lisa Kaczmarczyk

Student Team
Vincent Fiorentini, Hana Kim, Josh Petrack, and Hannah Young

Identification of missing persons becomes more difficult over time as the person ages. To address this challenge, this Clinic project aims to digitally age a face image of a person, building off of existing image processing techniques. The team has created a script interfacing with Adobe Photoshop to simulate the effects of aging on face images, focusing in particular on your subjects to better help identify missing children when their images are presented to automated tools and human observers.

Procedural Dynamic HDR Sky Rendering in Video Games

Client
Microsoft – Turn 10 Studios

Faculty Advisor
Elizabeth Sweedyk

Student Team
Melissa Galonsky, Karen Huddleston, Alejandro Mendoza, and Anna Pinson

Skies form an important part of any outdoor environment. Turn 10's current methods have to make tradeoffs between realism, disk space usage, and dynamically choosing the sun and cloud positions. The Turn 10 clinic team aims to solve these problems by procedurally generating realistic skies with the sun position and clouds asked for by the user.

Project CharlieMike

Client
Oakley

Faculty Advisor
Ran Libeskind-Hadas

Student Team
Marina Haukness, Bryan Mehall, Hannah Rose, Oliver Seifert, and Ashuka Xue

Joint project with the engineering department

Park of Oakley's design process requires the alignment of pairs of 3D scans of a human face and detection of key points of interest in these scans. Until now, this process was done manually and took approximately an hour for one pair of scans. The Oakley clinic team has developed software that automates the alignment and key point detection processes. This software substantially speeds up these tasks and provides more accurate results tan were previously attainable.

Tracking Activity and Motion to Improve Success Likelihood of Liver Transplant

Client
Project Spock

Faculty Advisor
David Harris

Student Team
Akhil Bagaria, Samantha Echevarria, Erin Paeng, Charlotte Robinson, Vaibhav Viswanathan, and Minhtrang Vy

Joint project with the engineering department

Several factors drive liver transplant success, including a patient's physical activity both pre- and post-operation, and the treatment of subsequent neurological complications that manifest as hand tremors. The Project Spock team is developing both an app for a wearable device and a secure web app that will allow physicians to track physical activity and detect the onset of hand tremors in their patients. These apps will help physicians monitor patients as they prepare for and recover from a liver transplant.

Advanced Toolkit for Adaptive Particle Simulations

Client
Sandia National Laboratory

Faculty Advisor
Jeff Amelang

Student Team
Daniel Bork, Maxfield Comstock, Justin Lee, and Matthew Valentine

The objective of the clinic project is to develop an alternative to a Voronoi tessellation package called voro++ to be used for mesh-free particle methods calculations. This includes efficiently matching existing functionality for identifying neighbors and calculating particle volumes. On top of this, the project will be able to support group specifications to be able to run calculations on subsets of the particles. The tesselator will interface with the mesh-free particle method package created by Sandia called MOAB.

Wood Veneer Classification

Client
Steelcase, Inc.

Faculty Advisor
Jim Boerkoel

Student Team
Jenner Felton, Perry Holen, Jennifer Rogers, and Rachel Wilson

Steelcase is the world's leading manufacturer of high-quality office furniture, some of which is finished with thin slices of wood, called veneer. Since wood is a natural product, this veneer may look significantly different from one log to the next. Our goal is to maximize customer satisfaction by minimizing the perceived variation among the pieces of veneer in furniture. To accomplish this goal, we built a model that sorts veneer into categories based on its color and grain characteristics.

Automated Examination of Collectibles

Client
TAG, LLC

Faculty Advisor
Zachary Dodds

Student Team
Kate Aplin, Hayden Blauzvern, Megan Shao, Ben Teng, and Avi Thaker

Joint project with the engineering department

Often, the examination of objects where condition is important do not use technology to its full capacity. Currently, these examinations are done by hand and by eye. Though traditional, this process I more costly, inconsistent, and generally not reproducible. To solve this problem our team has investigated and prototyped an algorithmically driven processing system wherein specific criteria are examined using computer vision and machine learning resulting in precise, consistent and reproducible conclusions.

Image Fingerprinting

Client
Time, Inc.

Faculty Advisor
Yi-Chieh (Jessica) Wu

Student Team
Jonathan Finnell, Sarah Gilkinson, Hannah Turk, and Zehao Zhang

Time, Inc. uses millions of images for its various brand websites. It is important for Time, Inc. to maximize reuse of content across brands while preventing perceived duplication and to track usage of images from various sources even after images have been resized, compressed, or other wise altered. Using content-based image hashing techniques, our team has created a system to identify similar images even after they have gone through editing processes.

Social Reputation on Twitter

Client
Webroot, Inc.

Faculty Advisor
Lisa Kaczmarczyk

Student Team
Brenda Garcia, Yuwei He, Ruoxi Lu, and Kevin Wynn

The goal for the Webroot 2015-2016 Clinic project is to increase the awareness of the risks users face when choosing to participate in various social media networks. The team has developed a social network reputation model, which will be available in the form of an Android application, to inform users of these risks. For this project, Twitter is the primary social network platform used for an analysis of the wide range of possible malicious content such as viruses and spam.

Empowering Information Discovery: Extracting Meaning from Yelp Reviews

Client
Yelp, Inc.

Faculty Advisor
Julie Medero

Student Team
Chloé Calvarin, Dani Demas, Josh Kutsko, Won Kyoung Park, and James Saindon

The goal of this project is to explore new ways in which Yelp can help its users glean information from an overwhelming number of reviews for a business. To that end, we have implemented a system that automatically identifies important characteristics of a business and the reviewers' opinions of these characteristics. With this information, we group these characteristics into high-level categories so that users can quickly find the information they're looking for.

Zenscript

Client
Zendesk, Inc.

Faculty Advisor
Ben Wiedermann

Student Team
Mai Ho, Shannon Lin, Kevin McSwiggen, and Alexander Putman

The Zendesk clinic team is developing a domain specific language to suit the needs of Zendesk's largest customers. The language is intended to provide a way to create more succinct and powerful business rules for defining customer support workflow than currently provided in Zendesk. The team's goal is to make these business rules powerful in order to reduce their overall quantity, improving the maintainability of the customer's system.