I am a talented problem solver and excel at accomplishing major tasks with extremely limited resources. I've got a bachelor's and a master's degree in electrical engineering, am interested in everything from machine learning to human perception/cognition to design to photography to music performance, and most likely know more than a little bit about your area of interest.
If you're interested in working with me to put my wide variety of skills together in order build a big cool thing, give me a ring or shoot me an email.
employment history
Computer Vision/Machine Learning Scientist at Geovisual Analytics—3 years
I develop tools and techniques to assess the state of agricultural fields containing specialty crops at significant scale. In the process of doing so I solve multi-domain infrastructure challenges. The multi-spectral imaging systems I built have collected many thousands of acres of images with 1-centimeter pixels.
- Imaging System Design and Integration
- Conducted field studies to assess the business and analytical viability of imaging sensors.
- Led project to develop mobile, nimble, and efficient agricultural imaging system
- Conducted design, integration, and test of camera system, drone, and inertial measurement unit
- Deployed three copies of the sensor package for ongoing data collection operations
- GIS and Imaging Pipeline
- Designed and implemented a fully-automatic direct georeferencing workflow to facilitate orthorectified image production with centimeter-level accuracy
- Designed and built a system that reconciles georeferenced images with known areas of interest and crops/combines them to create an image in a suitable format for analysis.
- Contributed to a GIS open-source project.
- Packaged tools in containers for simplified deployment.
- Computer Vision/Machine Learning
- Built a neural net-based crop classifier using time series images from satellite imagery that distinguishes among 5 vegetable crops with ~91% overall accuracy.
- Built tools to classify image content and gather information suitable to infer operational insights.
- Conducted experiments to determine the suitability of common machine learning techniques for extracting actionable data from images.
- Wrote Python bindings to utilize CUDA capabilities in OpenCV.
Expert at the Institute for Telecommunication Sciences—2 years
I design and implement no-reference speech quality models. These models use convolutional neural nets and knowledge about speech processing and impairments to achieve prediction-truth correlations >0.91, outperforming state-of-the-art models.
Electronics Engineer at the Institute for Telecommunication Sciences—10 years
I designed and conducted experiments that answer unique questions about how human perception of aural and visual stimulus correlates to the performance of communication systems. I then analyzed, documented and presented the results in the appropriate conference or journal. Whether officially representing the U.S. Government to international standards bodies, providing consult in one of my areas of expertise, building and maintaining infrastructure, developing processes or building an entirely new research capability, I found the opportunity to solve challenging problems extremely rewarding.
- Experiment Design, Data Analysis and Reporting
- Co-designed 2 speech intelligibility experiments and have administered, analyzed, and reported on 4 others
- Represented the U.S. Department of Commerce by presenting two reports [first, second] and informed codec selection activities for mission-critical push-to-talk functionality in 3GPP Release 13 LTE standards.
- This effort was recognized by former Secretary of Commerce Penny Pritzker with a U.S. Department of Commerce Silver Medal Award.
- Designed an experiment to collect performance data and thoroughly compare H.264 and H.265 performance and assess the suitability of common objective quality metrics to stand in for human perception experiments
- The results show that H.265's efficiency gains vary significantly depending on content type and that PSNR and even VQM are not highly correlated with subjective human quality assessments.
- Designed an experiment to measure the impact of non-standard viewing environments and non-standard monitors (including laptops and mobile devices) on human perception of video quality
- This experiment demonstrated that good/excellent video quality can be achieved without delivering full-resolution video to mobile devices and that further there may be no benefit to doing so.
- Helped build an experiment designed to assess whether and under what conditions crowdsourcing can augment and/or replace highly controlled intelligibility experiments
- Designed an experiment to measure a listener's ability to recognize who is talking given a system under test
- Completed literature search, experiment design, UI development, implementation, data analysis, documentation and publication in 9 months as the primary researcher during my first year of employ.
- These experiments (and others not listed) resulted in 20 peer-reviewed publications
- Project Management
- Lead a research project investigating radio frequency interactions among IoT radio access technologies through repeatable experiments to measure spectrum usage and interaction.
- Lead a project to build modular and extensible signal storage and processing engine to store and automate processing of audio/video/signal libraries.
- Programming
- Developed Python programs to add noise to, normalize, resample and transcode massive audio stimulus libraries among arbitrary codecs.
- Wrote front-end and back-end components to support a major crowdsourcing initiative including a custom interface for audio playback, resource download checking, filename obfuscation and automatic interface code generation.
- Designed a full-stack, multi-platform audio/video quality assessment platform enabling researchers to implement human subjective opinion measurement experiments using configuration files.
- Ported iLBC speech codec from C to Matlab to facilitate signal processing research.
- Wrote extensive algorithmic and signal processing programs in Matlab.
- Developed experimental UI in Matlab for use in novel human subjective opinion measurement experiments.
- Sysadmin
- Maintained public facing web and FTP servers that serve as a video library distribution system as well as an internal Gitlab instance that motivated use of source code control among engineers
- Awards
- United States Department of Commerce Silver Medal Award for Scientific and Engineering Achievement, 2016
- Elevation to Senior Member of the Institute of Electrical and Electronics Engineers in 2013.
- ITS Outstanding Publication Award, “Multiple-Description Speech Coding using Speech-Polarity Decomposition,” 2010
- EFPL Best Paper Award given at the First International Workshop on Quality of Multimedia Experience, “Gradient Ascent Paired-Comparison Subjective Quality Testing,” 2009
- ITS Outstanding Publication Award, “Intelligibility of Selected Radio Systems in the Presence of Fireground Noise,” 2008