As a Machine Learning Engineer on the Anti-Abuse team I work on the development of models for the identification of fraudulent job posting activity. I was responsible for creating and shipping the very first jobs fraud model that resulted in a 58% reduction in the number of applications to fraudulent job postings. I've also worked on several analyses that have led to major shifts in our team's quarterly focus as well as the discovery of one issue in particular that resulted in a 50% reduction in the work load of our human reviewers.

I am the team lead for the group responsible for the creation of the Applicant Tracking System (ATS) for LinkedIn Engineering. My team operates like a startup inside of LinkedIn. We use Agile methods and a tech stack geared towards a fast moving, interactive development cycle. On this team I serve many roles such as that of the technical lead responsible for ensuring the quality and future vision of our software. I also act as a mentor to many of the more junior members of the team. I am also the Python expert for the team where I am always available to answer questions about our tech stack or help out with the design of a feature. Finally, as part of my duties as the tech lead, I've tried to push us towards a more test driven development strategy. To facilitate this approach to development, I created the open source test library Flask-Fixtures (https://pypi.python.org/pypi/Flask-Fixtures).

Author of the title "Matplotlib for Developers: Data Visualization and Analysis with Python".

I contributed articles and screencasts for several Envato websites on technologies such as Node.js, Python, and Django to name a few. Examples of my work can be found on my Tuts+ author's page at http://tutsplus.com/authors/christopher-roach.

I created a framework (http://pypi.python.org/pypi/django-simple-rest/) for building RESTful APIs in Django which I then used to create the API that drives the Freshplum service. I also created the iOS library that's used to integrate iPhone/iPad applications with the Freshplum service. I used Python, Matplotlib, R, and Hadoop (AWS EMR) as well as several other technologies to explore and create products around the data that Freshplum collects. For example, I worked on one project where I designed and implemented an anomaly detection algorithm that would automatically uncover "interesting" stories in the data for our customers. As the first hire at Freshplum, my job description could most easily be summed up as "anything that needs to be done".

During my time at Apple I worked on two different teams. I spent several years working on the Online Store team where I, and a small team of coworkers, were responsible for completely redesigning and rewriting the multi-billion dollar cash register that is responsible for processing all of Apple's online sales. This accounts for both the front-end and back-end portions of Apple's online shopping cart and also powers other Apple apps such as the shopping cart portion of the Garage Band application. I also worked on several other aspects of the Online Store in addition to the shopping cart portion.In my final year at Apple, I transitioned to the team responsible for the Apple Store iPhone application. I used my training in Objective-C, Cocoa, and Cocoa Touch/iOS technologies to help in the creation and maintenance of new features to the application. During my team on this team, we launched several new features including a worldwide push.

At Harris, I worked on several projects from low-level embedded software engineering to customer-facing GUI development. On my first project I was in charge of creating the software that was used to control the modem for a small antenna array. I was given quite a bit of personal freedom in coming up with the design of the software and communication mechanism, so I went with a standard TCP socket communication and designed the communication protocol myself. The software was written in C# and essentially gave the user a graphical means to control the array.In another project, I took over the lead position for the software that would drive a SATCOM modem. For this project, the customer was familiar with, and wanted to use Matlab, to control the modem, so I used Python (Jython) and Java to write an intermediate layer that would allow Matlab to communicate with, and control, our modem over an RS-232 serial interface. The embedded software for the modem had a soft-real time requirement, and so we were able to use C# for its development. At the time I was teaching a class on C# at Harris and two of the students in my class were the Chief Software Engineer (my boss) for the project and another senior teammate. As a result, while I did not work directly on the embedded software for this project, one of my main roles was as a C# consultant to the engineers working on the embedded software for the modem.

At ACI, I worked on several data driven websites using mainly Microsoft technologies such as ASP and .NET, one of which was the personal website of Virginia Tech Hokies football coach Frank Beamer (http://beamerball.com). During my time at ACI, I was a proponent of, and pushed for, the use of open source software. As a result, I was given the chance to develop the company's first commercial project using LAMP technologies such as PostgreSQL, Linux, and Perl.