<a href="https://www.cmswire.com/open-source-cms/the-consequences-of-a-changing-open-source-software-business-model/" target="_blank">The Consequences of a Changing Open-Source Software Business Model</a>&nbsp;&nbsp;<font color="#6f6f6f">CMSWire</font>

The Consequences of a Changing Open-Source Software Business Model - CMSWire

The George Washington University

Co-founder & CTO of Docket™

Docket® partners with public health agencies and healthcare providers to expand access to personal health records, cut administrative overhead, and advance patient engagement through intuitive digital experiences. In 2018, Docket® teamed-up with HHS' Office of the CTO and Audacious Inquiry to expand consumer access to personal and family immunization records - live in Utah with additional state roll-outs planned through 2022.

First Place - U.S. Department of Health and Human Services' (HHS) Move Health Data Forward Challenge: http://www.hhs.gov/about/news/2016/11/02/hhs-announces-phase-1-winners-move-health-data-forward-challenge.html

Foxhall Wythe

Co-Founder and CTO

Greenfield AWS Training and Certification project.

Amazon Web Services (AWS)

Senior Software Engineer

Confidential greenfield project. Hand-picked to join a team of experienced Amazon senior SDEs to lead a major new business initiative. Resolved major, broad, ambiguous technical questions ("How do we get X piece of data?") into comprehensive high-level designs. Led a team of 6 engineers to build and orchestrate a fundamental service for this new business. Convinced other stakeholders on correct technical direction.

AWS Time Sync, EC2 Core

Time Sync owns all of AWS's NTP infrastructure. The team owns everything from the physical hardware in the data centers to the software API interface in a customer's EC2 instance.

I started on the team while it was located in Seattle. Worked with another senior SDE to transition to the team to HQ2 in Arlington where I anchored a rebuild of the team, including a new manager. I was responsible for training all new hires and creating effective onboarding material. This included preparing new engineers for on-call rotations for a major core tier-1 EC2 service.

There were several designs I wrote or influenced on this team, including a design to fix a major capability we had promised customers years before but never built and IPv6 changes that were required to support NTP.

AWS Lambda

The first major project I tackled on Lambda was re-designing the internal routing scheme as part of a larger service re-write. The routing layer was successfully launched, leading to an additional 3 9s of availability during Lambda's deployments. This improved routing scheme led to deterministic deployments, speeding up troubleshooting.

The second major project focused around internal encryption that would bring in business from clients with sensitive workloads. I wrote the design and led a team of 3-5 engineers (people came on and off the project as needed) to deliver this new internal capability. The work included finishing and fixing an internal encryption protocol, a high-performance Go server (requires were to add < 5ms latency at p99, with the crypto overhead included; can't use more than 10mb of memory to handle 126mb of payloads), and a pre-shared key infrastructure library that leveraged SigV4.

Software Engineer

SDE I/II

I work on the robot mitigation team which exists inside of the retail websites (e.g. amazon.com, amazon.co.uk, etc). We work to protect Amazon's retail fleet world-wide from bots and scrapers. Our objectives are to reduce infrastructure to save cost, while minimizing customer impact from mitigation actions like captchas.

My work is mostly in Java, making extensive use of Amazon Web Services, especially Redshift.

Specifically, I work on improving instrumentation, false positive rates, and verification of traffic from trusted 3rd parties. This includes linking together data across 6 teams, modifying critical code paths that every request to the retail website goes through, and actively monitoring and supporting those pieces. This has led to improved detection, with a final accuracy rate of around 6 9s.

I also re-designed the way we store and query data. We were using a few large Redshift clusters to store all the data we needed, but we were running out of I/O. I wrote up a design where our clusters were sharded - at the cluster level - to work around I/O overhead, and then implemented it with the help of my team. This also gave us automatic snapshotting and restoration for historic investigations into data.

Additionally I hunt down major traffic anomalies. This includes physically finding a source based on geodata, decompiling software then tracing their call patterns, and thwarting a bot's business case.

I've been listed as an inventor on three patent applications so far, two as the primary.

I mentor several other developers across the company, which I greatly enjoy. Additionally I started a bi-monthly best-practices review so the team stays sharp and manages tech debt.

Amazon

Software Developer

RJMetrics is a business intelligence company that allows you to aggregate all of your different data sources into a common area where a smarter, easier to digest analysis can be done. I worked extensively with Clojure, nodejs / coffeescript (server-side, not client-side), and PHP.

Most of my work centered around two areas: performance improvement for externally facing APIs, and moving the codebase from a monolothic app to a service-oriented architecture.

RJM allows customers to hit a REST API to upload data directly if they want. This API is written in coffeescript with the Q promises library. There were a few performance bottlenecks that were causing intermittent problems. This led me to not only finding them, but re-writing several portions of the core code base, and filing a bug with Amazon's AWS client libraries. I reduced average latency by 66% and the occurrence of hanging requests (taking 60+ seconds that would time out) dropped to almost zero.

Moving to an SOA was my other major project. All work was done in Clojure. I helped write a library that allows other teams to build out RESTful services very quickly, as well as converting parts of the existing monolithic app to their new service counterpart. This involved figuring out ways that services can talk to one another, automatic service discovery, and a common interface between all the services.

RJMetrics

The flagship product is called LeanSentry. It's a monitoring service for Windows Server that is composed of a remote monitor as well as an optional agent service. I wrote large portions of both the agent and monitoring service to retrieve better data without impacting performance on a client's machine, including a very low overhead CPU profiler and SQL query capturing service.

In addition I designed and built large-scale performance frameworks that are used internally. This included pushing very large data streams to both the Amazon and Windows Azure cloud. I built a testing infrastructure in AWS that allowed us to remotely spin up as many machines as we needed in whatever installation configuration we chose. The machines would verify that the installation succeeded or failed, and we could also use them to generate large data streams to test on.

I also built out several parts of the front-facing web dashboard using Razor MVC. This included core features such as account sharing, promotional programs, web analytics, reporting services, and major billing system adjustments.

I also wrote a chart image generator using PhantomJS that would capture Highcharts output and embed them in emails.

Key skills include C#, Javascript/HTML/CSS/Razor MVC, PHP, MSSQL, Azure web and worker roles, the full AWS Suite, nuget package creation and management, and IIS management.

LeanServer

Work involved low-level C programming and boot loader modification.

Naval Research Laboratory

STEP Intern - CHACS

I performed robotics research under Dr. Rahul Simha at the George Washington University with money from the National Science Foundation. The goal was to make a unified robotics operating system, similar to ROS, but exclusively in Java. We wanted to have an API that freshmen could engage with, and upperclassmen could peel back to do more advanced behavior. We used 4 different robots and a simulator. You could write a program that interfaced with our API that then worked on all platforms.

Nathan Scott