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 2012-2013

Podium: Making Mobile Apps More Social

Client
Amazon in OC

Faculty Advisor
Z Sweedyk

Student Team
Erin Coughlan (PM), Meredith Murphy, Haakon Saxberg, and Michael Stock-Matthews
Amazon in OC would like to improve the customer experience with mobile apps on the Kindle Fire and other Android devices by adding in-app commenting. The HMC clinic team has developed an API that allows mobile app developers to easily integrate with social networks. Customers can now read comments left by their friends while running the app -at a line in a book, a scene in a movie, or a level in a game--as well as leave their own comments.

Automated Candidate Sourcing

Client
Amazon.com

Faculty Advisor
Melissa O’Neill

Student Team
Zachary Purdy (PM), Rai Feren, Joseph Dearden, Josh Oratz
Amazon.com needs to hire talented software engineers to maintain and develop their online software services. The industry-standard processes for finding candidates miss many strong developers and are time-consuming for Amazon's recruiters. The goal of this project is to create a tool that improves the efficiency and breadth of Amazon's candidate search by intelligently mining online resources.

Automatic Fission Track Identification and Analysis in Apatite Crystals

Client
Apatite to Zircon, Inc

Faculty Advisor
Christopher A. Stone

Student Team
Mary Rachel Stimson (PM), Eli Omernick (POM), Colin Bundschu, Peter Andrien
Apatite to Zircon, Inc. is a geological services firm that provides services analyzing grains of apatite crystals from rocks. Previously, apatite grain analysis had been completed by hand. Our team researched and improved upon existing automation techniques by using a number of image-processing and machine-learning algorithms. These automation techniques will make it easier for Apatite to Zircon, Inc. to analyze apatite grains and to educate new geologists.

Optimizing SSD Performance for OneFS Workloads

Client
EMC Isilon Storage Division

Faculty Advisor
Geoff Kuenning

Student Team
David Ersek, Brad Jensen (PM-S), Stephen Levine (PM-F), Matthew Prince
Isilon recently began offering storage nodes that incorporate solid-state disks (SSDs). Although SSDs differ significantly from hard drives, they employ a flash translation layer that allows them to be treated as hard drives. However, SSDs still have very different optimal access patterns. We developed methods for optimizing writes to SSDs in Isilon's OneFS file system.

Semi-Automated C++ Code Transformations

Client
Intel Corporation / Technion University

Faculty Advisor
Ben Widermann, HMC and Dr. Ayal Zaks, Technion University

Student Team
Jessi Peck (PM), Jeff Hemphill, Gary Lent (POM), Gal Dalal (Technion), Nir Greshler (Technion) and Gal Greshler (Technion)
Intel Corp. designs specialized hardware modules to repeatedly execute data processing algorithms. These algorithms are initially written and tested in C++. Experts then develop descriptions of hardware modules using SystemC (a C++ library for describing and simulating hardware) using the algorithms as reference. The transformation from general C++ to SystemC is slow and repetitive, so we designed a tool that automates parts of the process.

Mobile Document Image Processing

Client
Laserfiche

Faculty Advisor
Z Sweedyk

Student Team
Jason Yu (PM), Jordan Ezzell, Nick Kinkade, Ben Normoyle
Laserfiche provides content management solutions that include software to digitize and process documents. One capability of Laserfiche's software is to process images of documents taken with a Smartphone camera so they look as if they were scanned. Our goal is to improve the algorithm in this process so that background artifacts and border noise are removed and document text is straightened.

An Open-Source System for Malware Detection

Client
Proofpoint, Inc.

Faculty Advisor
Geoff Kuenning

Student Team
My Diem Ho (PM), Josiah Gaskin, Alex Eng, Lawrence Thibodeaux
Proofpoint, Inc. offers Security-as-a-Service to companies of medium to large size and is dedicated to protecting sensitive information from outside threats such as malware. They asked our Clinic team to develop a system to detect attacks carried in innocent-seeming files. We constructed a system from existing open-source components that can detect suspicious Microsoft Office and Rich Text Format documents by analyzing files statically. The system examines the files for malformed content and structure, and flags suspicious material for more detailed analysis by a sandboxing system.

Augmented Reality Interior Decorator

Client
QUALCOMM

Faculty Advisor
Colleen Lewis

Student Team
Jessica Stringham (PM-S), Sam Konowitch (PM-F), Ari Schumer, Zewei Wang
Vuforia™ is a software platform developed by Qualcomm® which enables developers to integrate image recognition-based augmented reality (AR) in their mobile applications. Our task was to design and implement an application that demonstrates the features of Vuforia and potential uses of enables users to choose furniture from a physical catalog using Vuforia's cloud target image recognition, place and manipulate a 3D model in their environment using AR, and take and share photos of their arrangement with the user's social network.

GPU Data Compression and Analysis Library

Client
Quantcast

Faculty Advisor
Ran Libeskind-Hadas

Student Team
Scott Breyfogle, Andrew Carter, Dietrich Langenbach (PM-F), Jack Ma (PM-S)
Quantcast Corporation is interested in using GPUs to reduce workload on their CPUs by migrating computationally intensive tasks to the GPU. They have identified operations such as data compression and string search, which require significant CPU time on their system, as possible targets for this migration. The team designed and implemented a versatile GPU library that can execute these operations both on the CPU, and on the GPU using both CUDA and OpenCL frameworks. The team implemented and benchmarked six compression algorithms and a string search algorithm using the framework.

Development of an Advanced Finite Element Implementation for Atomistic-to-Continuum Modeling

Client
Sandia National Laboratories

Faculty Advisor
Robert Keller

Student Team
Jordan Librande (PM), Jacob Bandes-Storch, Jay Jonsson, Michael Morton
The Atomistic-to-Continuum software package (AtC), developed by Sandia National Laboratories, couples the modeling techniques of Molecular Dynamics and Finite Element Analysis in order to simulate a wide variety of complex nanoscale systems. Our goal was to increase the parallel performance of A5C, thereby making larger simulations more computationally feasible in multi-core environments. To this end, we designed and implemented parallel matrix multiplication algorithms and distributed Finite Element data and calculations across processors more intelligently.

A User Interface for VMware's Vprobes Software

Client
VMware, Inc.

Faculty Advisor
Christopher A. Stone

Student Team
Amelia Nybakke, Max Kukartsev (PM-S), Milo Toor, Sarah Johnson (PM-F)
VMware, Inc. is a leading developer of virtual machine technology. In order to diagnose software running within the virtual setting, VMware provides a script-based instrumentation technology called VProbes. However, this technology has a steep learning curve and requires the user to be familiar with the underlying system of VMware's software. In our Clinic project we made VProbes accessible to non-programmers by exposing some of its common functionality via a web-based user interface.

Interactive Documentation Collaboration System

Client
Walt Disney Animation Studios

Faculty Advisor
Robert Keller

Student Team
Beryl Egerter, Julius Elinson (PM), Jessica Hester, August Toman-Yih
Artists and developers at Walt Disney Animation Studios work with rapidly evolving internal software and complex commercial packages. Staying current with new features, best practices, and studio-specific workflows can be challenging. Users frequently have questions about software capabilities or discover aspects that would be useful to their colleagues. However, there formerly was no effective means of making inquiries or sharing such information. We have developed CPlane, a program that will enable users to create and share this documentation easily.