The Compute Platform group consists of Compute Infrastructure teams building, operating and maintaining Yelp's platform as a service (PaaSTA) running on k8s, and Production Engineering teams that own Reliability and Observability. My role as platform lead is to identify the strategies and opportunities for cross-team collaboration that align to our group's vision for teams to execute on, while delivering a reliable and cost efficient compute platform for workloads at Yelp.- Drive product vision to deliver a differentiated Compute Platform product that powers Yelp’s business.- Define and measure key performance indicators to quantify product value, prioritize initiatives, and build the business case for investment (build vs buy decisions for multiple new and existing vendor relationships, how to evaluate success, etc.).- Drive innovation through user research, staying up to date with industry trends, and building strong vendor / open-source relationships (e.g., State of FinOps Survey with FinOps Foundation).- Create and operate key AWS cloud financial management processes (e.g., reporting to executive stakeholders in and out of Engineering, forecasting, negotiating, and committed spend planning).

The Production Observability team empowers developers to understand and operate their systems in production; building and promoting standard and reusable approaches to use logs, metrics, traces, dashboards, alerts / SLOs in order to verify the site is available, fast and serving correct responses to users. The team builds tooling for tracking and optimizing AWS costs engineering-wide.- Build a hierarchical cost categorization system to provide varying granularity roll-ups for differing stakeholders' reporting needs; used in biweekly reporting for engineering teams, standard alerts on cost increases- Created an ongoing automated container rightsizer tool / process for Yelp's PaaS, based on allocation versus utilization, to eliminate waste on cpu-driven cluster autoscaling- Track and optimize the cost of Yelp's largest shared S3 buckets by prefix, and implement more granular lifecycle transition policies to shift data to lower-cost storage classes- Document and implement centrally managed convertible reservation strategy, with automated instance type conversions (via vendor tooling)- Inform Yelp's long-term observability needs by working closely with power users, soliciting developer feedback, attending industry events, etc.

The infrastructure performance team owns the performance of Yelp's websites and mobile applications. We create tools and dashboards to enable easily tracking Yelp's website and app performance, and we build and maintain the high performance of core infrastructure that features are built on.• Regenerated over 100 million images using Elastic MapReduce to apply best practice image compression techniques, saving an average 30% to image file size for all Yelp images without any loss in perceived quality. This resulted in a 15% reduction in CDN bandwidth. Led to a top 15 all-time viewed blog post on Yelp's engineering blog.• Built a system for load tests running Python with Locust.io and Docker, integrated with Jenkins, for opt-in load testing on merging branches.• Large-scale refactor to HAProxy and Nginx configurations to permit TCP connections to be reused for internal cross-datacenter HTTP requests, reducing east coast and international master data center response times by 60ms.• Ad-hoc analyses for production investigation and regression using Splunk, Signalfx, profiling and distributed tracing tooling (Zipkin).• Consider various optimizations to Yelp's core infrastructure based on performance analysis data weighing trade-offs in reductions to user-facing response time, reliability, resilience, and infrastructure cost.

Various positions for clients consulting on IT performance -- including load testing, defect and incident analysis, monitoring, tuning.• Monitored and built dashboards via tools like Sitescope and Splunk to watch infrastructure and application logs during tests.• Identified performance issues with tracing profilers (JProfiler, YourKit) and APM (AppDynamics).• Reviewed and changed source code to fix performance bottlenecks.• Created automation scripts for load testing using industry standard tools like LoadRunner, Performance Center, JMeter, SoapUI.• Deployed and integrated tools like WebPageTest and Sitespeed.io into Jenkins for automated performance regression testing.• Configured AWS EC2 instances to scale horizontally, in order to meet additional infrastructure needs of load generation.

Used a variety of tools to get the job done across several projects.Java, .NET, SQL Server, Informatica, DB2, UnixSupported production team on federal government system processing millions of health insurance claims per day through InformaticaWorked across the stack on a Java 6 web applicationDeveloped .NET winform warehouse POS component from requirementsDocumented BI requirementsCreated an SSRS Custom Security ExtensionManaged a 12-member team responsible for prototyping a few hundred ASP.NET pages to high-level spec