professional summary
For the last ten years, I’ve specialized in designing and building analytic software systems, using techniques from artificial intelligence, machine learning, statistical analysis, computational linguistics, and pattern recognition.
I’ve analyzed all sorts of data, looking for patterns in historic stock market prices, or for clusters of like-minded movie watchers in a dataset with a hundred million ratings. I’ve contributed to criminal investigations by finding correlations in electronic forensic data, and I’ve identified failure modes of mechanical systems by parsing the text in automotive warranty databases. I’ve modeled missile defense simulations and studied heat emissivity patterns in the vascular structures of the human face.
In each of these scenarios, I’ve developed algorithms either to identify the emergent aggregate characteristics of the dataset, or to extract statistically significant kernels of information from among hundreds of millions of documents. I usually work with domain experts – soaking up their knowledge and expertise – to transform their methodologies and hypotheses into concrete software implementations.
I'm happiest in a small company environment where I can perform research and build prototypes, working directly on products while thinking strategically to meet clients’ business objectives. I thrive in a collaborative setting, brainstorming with coworkers and showing leadership by building consensus among stakeholders for new ideas and initiatives.
recent work experience
Software Engineer, Charles River Analytics, October 2008 to Present
Charles River Analytics is a defense department contractor working on cutting edge R&D projects for the Army, Navy, Air Force, and DARPA. In the “Cognitive Systems” division, focused on modeling, comprehension, and augmentation of human cognition in a military context. Responsibilities include grant proposal authorship, project management, travel for client presentations, and technical implementation.
Accomplishments:
- Wrote a winning Navy SBIR grant proposal to fund preliminary research in infrared image analysis of the human face for the classification of stress, anxiety, uncertainty, and fatigue. Developed algorithms for the extraction of statistically significant features from the visible and long-wave thermal infrared spectra. Designed an experimental protocol for data collection and hypothesis validation. Managed relationships with our Navy client as well as with a collaborative implementation team at the University of Massachusetts.
- Served as Technical Lead on a $600K research project for the Missile Defense Agency. Implemented a simulation system using Monte Carlo methods to evaluate tens of thousands of enemy missile attack scenarios and the relative efficacy of various RADAR defense configurations. Deployed a 3D geospatial visualization system rending all simulation results as a volumetric cloud, enabling strategic RADAR planners to perform exploratory analysis of simulation results.
- Developed a geospatial “heat-map” visualization component based on Gaussian kernel density estimators to provide accurate interpolative renderings of measurements over a two-dimensional field. Profiled and optimized the component to handle thousands of independent measurements using a cluster-based approach to grouping and aggregating the measurements.
Consultant, Wry Research, November 2007 to October 2008
Developed and delivered custom software solutions to clients in aerospace/defense, medical research, and new media industries. Assessed client needs, defined project requirements, planned project schedules, and provided regular status reports to key stakeholders.
Client/Project Highlights:
- Raytheon Corp: Simulated massive-scale multi-user collaborative behavior to test client’s critical insider-threat detection software, instrumenting and automating a network of physical workstations to enact scenarios of data theft and miscompliance. Developed an agent-based simulation system capable of simultaneously processing the behavioral activities and interactions between tens of thousands of independent agents.
- Columbia University Medical School: Launched a study of neurosurgeon decision-making in aneurism cases, focused on quantifying the contribution of each patient variable toward a final treatment decision and finding the primary contributors of “disconsensus” among groups of physicians. Developed a multidimensional information density model, using sparse regions of the vector-space to inform the solicitation of new data in a “crowdsourcing” environment.
Senior Scientist, MatchMine, March 2007 to November 2007
MatchMine developed a vendor-neutral platform for content discovery. Much like Amazon or Netflix compute recommendations from the purchase and rental histories of their users, MatchMine facilitates the delivery of recommendations from third-party content providers to their customers, based on a statistical preference model of the individual user, and of the user community at large.
Accomplishments:
- Proposed a statistical behavioral framework (using vector-space mathematics) to model the multidimensional attributes of our core recommendation system.
- Hired a small team of research scientists, with specializations in computational linguistics and statistical information modeling, to flesh out the implementation details of the core algorithmic model.
- Developed a suite of prototyping and testing tools for the development and enhancement of core algorithms, including functionality for analyzing density distributions and vector magnitude histograms within the multidimensional space, and for performing similarity analysis between content items, users, user groups, and user archetypes.
- Designed and implemented a multi-threaded, context-directed web crawler –with the ability to predict link-target document quality – to fulfill the company’s content acquisition strategy.
Research Team Lead, Oakley Networks, July 2005 to March 2007
Oakley Networks develops specialized network security software for managing insider threat risks. Oakley’s software enables its users to create policies restricting the transfer of sensitive intellectual property on the network, and to take action upon detection of policy breaches.
Accomplishments:
- Managed a team of developers working on a variety of research and analytical projects, including compiler tools, classification engines, indexing tools, and rules-based logic engines.
- Developed a domain-specific language – and implemented the compiler – for Oakley’s security software. Language features included: static type-checking, namespace-aware symbol resolution, multiple inheritance, abstract classes, anonymous inner classes, integrated unit tests, and API doc generation.
- Proposed and developed a novel file-type identification algorithm using supervised learning techniques, including byte frequency indexing, agglomerative clustering, and probabilistic index reduction to produce fast and accurate results from a set of tiny “fingerprint” files.
- Developed a new technique for conducting massively parallel search operations through transient text documents using deterministic finite automata. This technique enabled the software to perform tens of thousands of simultaneous search operations per document in a fraction of the time previously required.
Computational Linguist, Attensity Corp, December 2002 to March 2005
Attensity's linguistic software reads English-language documents, producing parse trees from the source text and performing transformations on those trees to enable complex analysis of the entities, events, and relationships discussed within the text.
Accomplishments:
- Served as technical lead on a $250K pilot project for a large government account, including the maintenance of the primary customer relationship. During the project's three-month timeline, my four-man team delivered an implementation that resulted in over $1.5M in licensing revenue and an additional $1M in service contracts.
- Hand-picked to lead a turnaround team on critical project, a three-month assignment to improve the linguistic classification accuracy of a major project for the company’s biggest client. At the conclusion of the project — with accuracy more than doubled — the client signed a $2M contract for an extension of services.
- Developed a four-day training course to teach members of the intelligence community, from various federal agencies, how to use Attensity software in their own linguistic analysis projects. Traveled to Washington DC on many occasions to deliver that course.
- Contributed to the company’s core linguistic IP, including the development of parsing heuristics, named entity extraction rules, and part-of-speech disambiguators.
languages & tools
Java • SQL • Flex • ActionScript • .NET • C# • WPF/XAML
Python • Perl • PHP • HTML • JavaScript • OpenGL • C++
Eclipse • Visual Studio .NET • Subversion • CVS • Apache Server
Tomcat • Jetty • Jira • Confluence • MySQL • MS SQL
Windows • MS Office • Linux • Cygwin
education
Coursework in Computer Science
University of Utah, Salt Lake City, Utah
Bachelor of Arts in Theatre/Playwriting
Brigham Young University, Provo, Utah