Core API Platform - GraphQLExecuted HomeTimeline migration from REST to GraphQL by analyzing, documenting, disseminating, and driving efficiency improvements to handle 45k+ fields per query, 100k’s of queries per second, resulting in 50% fewer compute and observability resources, saving $$ millions annuallyIdentified developer experience issues, educated team on best practices, enabled 8+ engineers to deliver improvements, yielding savings of hours per individual change, aggregating to quarters/years of saved engineering time and resources across the company

Engineer on the Core Systems Libraries (CSL) team, which include Twitter OSS projects - Finagle and Finatra.Streamlined Privacy and Data Protection compliance (GDPR, FTC Consent Decree) via design, implementation, documentation, and support of Finatra mTLS framework integrationDeveloper Experience - Saved quarters of engineering time and compute resources by reducing Finatra’s embedded server startup time by 99.5%Reduced HTTP connections established in datacenters 8x, thus stabilizing request tail latencies for mTLS. Achieved by driving to completion the 5+ year Finagle HTTP/2 client migration via implementation and gradual roll out of Flow Control support. I directly supported and enabled previously burned partner teams, empowering them to confidently own their migration

Twitter Observability TeamTech Lead for the mission critical Observability stack responsible for visualization, alerting, storage, and querying over 4.3 Billion unique operational service metrics written per minuteStabilized team, growing from 4 to 12 engineers, lead extensive knowledge sharing, drove 90% reduction of 24x7 on-call alerts through bolstering, optimizing, and rearchitecting distributed servicesExecuted and Advised Alerting System Migration by understanding user needs, design & implement features, document, and provide direct support to 100% complete migration and decommission/delete legacy system. Monitorama PDX 2018 conference keynote talk based upon learnings from this effort

I have been helping make Watson more accessible to the masses, through the Watson Developer Cloud and the Watson Natural Language Classifier service available via BlueMix. I have driven the adoption of Behavior Driven Development within our engineering team, driven the use and education of Dependency Injection, architected key components of our REST service, designed our automated Integration Test suite (which includes performance testing and accuracy validation), as well as metrics, monitoring, and logging service tiers.

Software Performance Test Team Lead for Watson Solutions Core Division. Responsible for memory reduction, bottleneck discovery, code level optimizations, software architecture design, system measurement, analysis, and improvement. Designed and maintain scalable performance framework and tools for generating and measuring load against the Watson Pipeline. My day typically involves performance measurements, analyzing data, profiling (CPU, Memory, Garbage Collection), tuning the application and system, and developing code level solutions to the complex system that is Watson (yes, the Watson that won Jeopardy!).

Developed and maintained highly scalable performance tools for measuring the LotusLive Meetings web conferencing application and its various real time features. This included coverage and user simulation for audio, video, application share, slide presentation, and chat, amongst others. My responsibilities included measuring, analyzing, and presenting data at an executive level, along with maintain hardware, network deployment, software installation and tuning - in both physical and virtualized environments. During critical situations I was directly involved with customers (even traveled over seas) and the operations teams responsible for deploying the application into production use. I also designed and delivered code level optimizations. A major focus was on dealing with performance issues on a globally deployed real-time collaboration application and simulating accurate user behavior to gather metrics that represented real world, global usage and not just unrealistic numbers gathered in a pristine lab environment.