capture from multiple automotive cameras. Each collection processed and saved multiple GB/s

over multiple hours with no data loss. Enabled on-road data collections, imperative to sensor

evaluation for future Cruise programs. Multi-threaded C++17; Solectrix hardware; Dell server.

camera and real-time statistics (e.g. pixel histogram). Massive performance improvements over

legacy, Python-based visualizer. Qt5, multi-threaded C++17.

drive video. Reported color statistics of each TL for disparity analysis. Detected TLs via

YOLOv4, built/run in separate Docker container and exposed via custom Python Flask API.

experimental features of chip to process and clone single-frame inputs, allowing for direct com-

parison of reference frame processing results from multiple ISP tunings in parallel.

I helped the Doma Comms team implement, test, debug, release, and support a web service that receives, intelligently classifies, processes, stores, and surfaces title/escrow emails to Doma’s Operations team. At the time of my departure, the service handled roughly 10,000 emails per business day (though the number was growing so fast that it may have been even larger, and certainly is now).

The service used Node, TypeScript, Express, React, and PostGres.

I was member of the PI Server team, which was responsible for making sure that data fed into OSI's PI System was stored safely, accurately, and with as small a memory/storage footprint as possible.

Small selection of relevant work:

Parallel implementation of 3D Growcut segmentation, used on MRI volumes. Data provided by ABIDE (http://fcon_1000.projects.nitrc.org/indi/abide/).

See attached link for video and press release.

I designed, implemented, and tested a mobile web application for collecting real-time audience input during the premiere of "Darest Thou Now, O Soul," composed by Dr. Steven Sametz. Input from the audience affected the display on a 10'x10'10' cube of 340 LED lights, hanging from the ceiling of the performance space.

There were about 2,000 present for the premiere: half were in Lehigh's Packer Memorial Chapel, where we performed, and half were watching a live video broadcast from Grace Hall. Both groups were able to use my app during the performance.

I wrote client-side JavaScript animations, designed a simple queue system to ensure relative synchrony across audience members' phones, and created a web server to both capture all audience input as well as to broadcast "next animation" messages to all phones and to the LED cube.