Building a unified global payments infrastructure at Primer.io.

Member of the executive team, reporting to the CEO. I was responsible for Forto's software engineering and Workplace Tech (aka "IT") organizations, focused on delivering innovative technology solutions that allow Forto to create unique value for our customers, partners, and stakeholders.

I ran a multi-hundred person organization, across 10 locations (Europe/Asia/N. and S. America), consisting of engineering, ML/AI, product management, and operations teams, working on a number of different aspects of Amazon's WW e-commerce business:1) The Product Localization team (see previous description): grew business impact further, to tens of billions of dollars in incremental revenue, by: scaling automated product cross-listing by 2.5x, thereby expanding Amazon's WW catalog by 25%; enabling multi-language support for third-party sellers on Amazon's US and Saudi Arabia marketplaces via real-time machine translation; and rationalizing business rules for cross-listing across all Amazon marketplaces. 2) A Search team responsible for improving product search functionality in 10 recently-launched Amazon e-commerce sites, such as Turkey, Egypt, and Poland. This was an entirely new ~50-person team built from scratch in under a year, on the basis of a business case constructed by my team, and consisted primarily of machine learning specialists. The team built and launched improvements to machine learning models targeting the relevance, comprehensiveness and click-through rate of search results in these countries, which resulted in up to 80% improvement in the associated metrics, and hundreds of millions of dollars in incremental revenue. 3) A Translation Services organization, which provided an Amazon-internal software and operations platform for translating content, via both human and machine translation, serving large internal customers like AWS, Alexa, and Kindle as they expanded internationally. Under my tenure, we moved the ~100 person team of human translators from Luxembourg to Spain and Slovakia, with no customer disruption, implemented automation that saved tens of headcount for our customers, and reduced yearly translation costs by ~1 million dollars.

I ran the Product Localization team, which is responsible for localizing product data across Amazon marketplace catalogues world-wide and drives multiple billions of dollars per year in incremental revenue. This team consisted of a ~200-person org distributed across Europe, India, Singapore, and South America. and included software engineering, machine learning, product management, and operational roles. During my tenure, I drove a fundamental service re-architecture which increased capacity by 12x while reducing latency by 3x and reducing unit costs by 85%. My team also launched Amazon's first neural-network based machine translation model in production, improved machine translation model training capacity by 4x, deployed tens of such models, and contributed to a paper published in the 2017 IWSLT machine learning conference. In addition, team morale (as measured by Amazon's internal employee survey) increased by 10%-25%, and was consistently above the company average.

I was part of the EC2 Control Plane team, which builds all the software required to provision EC2 instances and manage them throughout their life cycle. These control plane services run on all EC2 servers, and are the underpinning of all the EC2 APIs. My teams owned services involved in all EC2 instance creations, modifications, or terminations, and their contributions were critical to launching EC2's support for live migrations, migrating instances away from failing hardware, and supporting new instance types. Improvements made during my tenure also allowed us to smoothly reboot millions of EC2 hosts WW as part of security upgrades. My teams were also responsible for services used by AWS teams to replicate their service configuration across AZs/regions, and during my tenure launched features that grew the number of services utilizing these features by 6x, to ~1000.

I ran the engineering team for the AWS Relational Database Service (RDS), which is a cloud computing service that automates management of relational databases, and makes it easy to set up, operate, scale, and run a highly-available relational database (MySQL, Oracle, SQL Server, and PostgreSQL) in the cloud. I was responsible for multiple teams, consisting of a mix of development managers, SDEs, SDE-Ts, and Database Engineers. I grew the team by 3x; launched 40 new features, including a 6x increase in storage/throughput available from an RDS instance, a 90% reduction in failover time, and cross-region disaster recovery; and managed the service through a 10x growth in revenue and number of database instances being managed.

Development manager in the Microsoft adCenter team, managed 20-person development team responsible for building systems infrastructure for revenue and relevance algorithms used in determining which ads to show in paid search advertising scenarios.Team worked on offline and online systems for Paid Search advertising: offline data mining and machine learning pipelines; online stores and algorithms for ad selection, relevance evaluation, real-time auctions; experimentation infrastructure.Managed team through Yahoo! traffic and advertiser migration to Microsoft adCenter

Led team of developers working on the Windows Live Storage and Live Mesh services, a set of services designed to give customer seamless access to their data across all their devices, and in the cloud. My team was responsible for both customer-facing functionality, like client notifications and personalized search, as well as platform infrastructure pieces, like a generic distributed in-memory key-value store, a persistent queueing service, and monitoring. The systems we built handled thousands of requests/sec and had multi-hundred million document search indexes.

Responsible for overall product architecture of v1 of Microsoft’s Peer-to-Peer Software Development Toolkit.Managed team of 14 developers from design phase through shipping v1 of the product in July 2003.

Led team responsible for cluster management, membership and load-balancing functionality in Microsoft Application Center v1. Personally designed and implemented core clustering functionality, and load-balancing management features, with specific focus on integration with Network Load Balancing and Component Load Balancing.

Responsible for all security features of Internet Information Server.

I was part of the team that built Nynex Yellow Pages (BigYellow), an online yellow pages directory.