Andreas Girgensohn

Experienced full-stack software developer and software architect for web-based, iOS, and Android projects. Accomplished scientist able to leverage multimedia and human-computer interaction to create rich, usable multimedia applications resulting in many patents and publications. Technical leader, team player, and key contributor to many scientific and software development projects.

Technologies

Frameworks	React, Hooks, GraphQL, Apollo Client, Node.js, Express, Numpy, Flask, JAX-RS, .NET, Xamarin, Docker, AWS, MongoDB, DynamoDB, MySQL
Tools	JUnit, Jest, pytest, XCTest, valgrind, Plotly.js, D3, Matplotlib, Unity 3D, FFmpeg, Ghostscript
Languages	JavaScript, TypeScript, Java, Python, Swift, C++, C#, HTML, CSS
Systems	Ubuntu, Fedora, Raspberry Pi OS, Windows, macOS, iOS, Android
Deep learning	PyTorch, TensorFlow, TFLite

Employment

1996 - 2020	FX Palo Alto Laboratory (Fuji Xerox subsidiary), Distinguished Scientist (last three years; only one in the company; reported to CTO)
1992 - 1996	NYNEX Science and Technology (now Verizon), Member of Technical Staff
1987 - 1988	University of Stuttgart, Department of Computer Science, Research Associate

Education

1992	University of Colorado, Boulder, Computer Science, Ph.D.
1987	University of Stuttgart, Germany, Computer Science, M.S.

Additional Information

Portfolio of projects: https://girgensohn.com/andreas/portfolio/ (with videos of projects)
GitHub: https://github.com/andreasg123
LinkedIn: https://linkedin.com/in/andreas-girgensohn
61 Patents: https://girgensohn.com/andreas/patents.html (some pending)
104 Publications: https://girgensohn.com/andreas/publications.html
Google Scholar: https://scholar.google.com/citations?hl=en&user=Dz6O99EAAAAJ

Sample Technologies and Architectures Used in Projects (click titles for details)

3D face mesh. To estimate pain levels of medical patients, the iOS augmented reality kit was used to record a stream of 3D facial keypoints in a compact JSON format. Transferred the data via a mounted file system to use machine learning (TensorFlow). Displayed the face mesh stream in a WebGL canvas animated via JavaScript.
Computed tomography. Displayed DICOM CT images as a 3D image cube with synchronized slices. Used React to render page elements and an HTML canvas to display an image slice at the selected zoom and grayscale levels.
3D model of gastrointestinal tract. Used the Unity video game engine to automatically navigate through a model of the gastrointestinal tract while collecting images at different view angles. Determined image collection approaches best suited for the machine learning algorithms.
Activity forecasting. Created a Python wrapper for an open-source image library to provide the input for a deep learning model. Compared the GPU acceleration of different image processing approaches. Created a Docker container to simplify running an open-source GPU deep learning model.
Indoor localization. Implemented a particle-filter framework in C#, Java, and Python for mobile and desktop platforms to turn the readings from Bluetooth Low Energy and Wi-Fi RTT beacons into location estimates. Plotted the results onto floor plans drawn in Xamarin and Matplotlib.
Video surveillance. Recorded streams of IP surveillance cameras in a Java service providing web-based access to synchronized video recordings at different playback speeds, including reverse playback. Displayed live and recorded streams in a Java app using computer-vision techniques for stream selection and person tracking.
Copyright detection. Developed C++ image matching techniques and optimized them to fit into the memory constraints. Created a server architecture implemented in Java that stored data of copyrighted page images in MongoDB and called the image library via the Java Native Interface. Updated the architecture to use AWS Lambda and DynamoDB for image ingest, and Docker on AWS EC2 for periodic, resource-intense maintenance tasks. Monitored AWS costs to pick the most cost-effective approach.
Document browsing. Created a web-based app for browsing and searching a large collection of office documents. Developed a pipeline for converting documents to PDF via OpenOffice, extracting page images with Ghostscript, and scaling them to different sizes for a zoomable UI. Combined a Java web service with a JavaScript-rendered UI.
Social awareness. Created an Android app that reported its location in a battery-efficient way to a social awareness app. Created a year-long social network for a scientific conference using MySQL and Java Server Pages for a template-based creation of pages. Developed a social awareness system with frequently updated still images captured by video cameras at partipants’ desks.
Video conferencing. Combined WebRTC video conferencing and recorded video in a pipeline of MySQL, Node.js, WebSocket, client-side JavaScript, HTML, CSS. Displayed timelines in an HTML canvas. Incorporated an open-source WebRTC server as a multipoint control unit with video recording. Contributed to that open-source server written in C. Transcoded videos for playback with FFmpeg.
Video editing. Developed a video editor as a Java applet displaying images to represent video clips. Created a C++ DirectShow app to retrieve video from a DV camcorder, to analyze it for motion and other features, and to transcode it for playback. Developed another C++ app to produce the edited video.
Video and photo access. Created Java Swing apps for searching and organizing videos and photos. Used QuickTime for video playback inside the Java app. Sped up image loading by using a C JPEG library via the Java Native Interface.