As a member of a the data team, I worked on improve the data workflow of AI/ML engineers. The data command-line tool helps AI/ML engineers, and data scientists with data versioning, management and cloud transfer. The data is tracked locally, and transferred data in a docker registry for efficiency.
My first project on the team was to developed a way to record and visualize the tool’s CPU usage and memory consumption. I used Python (psutils) to capture the app's metrics, and create visualizations using Dash with Plotly for the results.
Later, I added the capability for the tool to give line specific feedback when users mistakenly adds invalid entry to configuration files (yaml, and json).
Currently, I am working on a tool to generate test datasets to exercises the features of the data catalog (an internal data sharing platform).
For my first project as a full-time developer, I created a prototype plug-in for Adobe Acrobat to increase the productivity of analysts processing redactions for multiple FOIA (Freedom of Information Act) requests.
Analysts process dozens of large PDF documents a week, and wet (ink) signatures require them to comb through a document page by page to redact them. I worked on a prototype application that enabled a set of features that would speed up their workflow. The notable features of the app are:
- Automated search of wet signatures within a PDF, previously a manual process, using OpenCV.
- A flexible redaction label manager. Instead of only applying one label to one redaction, the analyst can set labels to redaction in a one-to-one, one-to-many, many-to-one, and many-to-many in a single step.
While trying to achieve those results, I encountered a few problems. One of them was that signature lines led to many false negatives. So, I added a preprocessing layer to remove the lines before the search. The added layer reduced false negatives in search results by over 90% in cases where lines were present in the original selection. I also updated the app's architecture from single to multithreaded (along with the necessary data structures), which improved app UI responsiveness by pushing heavy computation to separate worker threads.
I led product sprint planning sessions by contributing issues and scoping features to implement within sprints.
Tools: C++11, Acrobat SDK, Visual Studio 2019, CMake
I designed, built, and tested an electromagnetic interference tolerant interface box using optical isolators and fiber-optic cables. The interface was used to obtain data for Federal Aviation Administration qualification procedures for the hardware we built. I also wrote and executed test procedures for the qualification of flight-ready electronic hardware and designed printed circuit board test adapters using Altium Designer to facilitate test procedures.
I researched, analyzed, and validated the protections devices we used on various circuits for the hardware we were producing at the time (Note: for the curious, look up transient suppression diodes). I used MATLAB to compute energy absorbed by the device and accounted for environmental factors such as temperature and altitude. I also analyzed the impact of a lightning strike on aircraft electrical sub-circuits.
I also wrote and managed the hardware requirements document for the 777X Backup Electrical Power System (BEPS) converter using DOORS.
I wrote and reviewed requirements for the aircraft health management unit. I also developed and executed test procedures for the aircraft health management unit.
- University of Dayton MSc in Electrical Engineering, Dayton OH, May 2017
- University of Notre Dame BSc in Electrical Engineering, Notre Dame, IN May 2014
- AWS Certified Solutions Architect – Associate April 2021
- Web Programming Technical Certificate, Sinclair Community College, Dayton, OH May 2020
- AWS Certified Developer – Associate Nov 2019