I lead 3 teams at Rho and several product areas.- Notifications. They are in charge of the email, SMS, and mobile notifications pipeline like 2fa, card alerts, daily digests, approval requests etc- Expenses team is in charge of the expense flow from when the card transaction was authorized until the expense was approved and synced with accounting. The system supports configurable expense rules, defines complex and flexible approval structures, and sends out various notifications to collect data and drive approvals forwards.- Authentication and Authorization team is in charge of everything auth-related. This includes a permissions structure and customer-definable user groups.

Keep production environment for Redfin.com running smoothly, performantly, with 99.9% availability.▪Debug production issues when they arise and decrease immediate impact.▪Build monitoring boards and alerts in Grafana, New Relic. ▪Identify potential bottlenecks beforehand by running controlled production attacks. These usually included increasing the load, CPU consumption and causing network saturation.▪Improve performance of various production services. For example: I debugged and resolved memory leaks in production Node servers. Optimised service to service communication on localhost by switching from connection per request model to unix domain socket. Debugged and performance tuned the Java API server, Tomcat and its front proxy Apache (the highest traffic production service) mostly using NewRelic, java async profiler's flame graphs and linux command line tools.▪Introduced and deployed Envoy as the internal service to service load balancer. The internal traffic used to go through the F5 BigIP which also handles external traffic and traffic to AWS. BigIP was becoming a bottleneck and a risky single point of failure. Envoy also enabled us to deploy detailed monitoring of services out of the box.

I was asked to lead the Experimentation team through the transition period to make sure that all ongoing projects are prioritised correctly and completed on time. After a period of 6 months the team was to split and join different orgs. ▪Understand the existing pipeline and prioritise ongoing project in order to achieve clear picture in 6 months timeframe. Experimentation pipeline is the top producer of data at Redfin producing millions of events per second. Data enters the pipeline through the NGINX plugin and a few server side endpoints from where it is sent to AWS for further processing and eventually being stored in Redshift. ▪Work with Analytics and Product Managers who are the primary users of the pipeline and make sure that their requirements are addressed. ▪Understand and address technical debt before the team is split.▪Remove unnecessary projects from the roadmap and shift engineering efforts where they are most needed.

▪I manage the team of 6 engineers which focuses on monitoring and improving the overall web site performance. Performance of Redfin.com is directly tied to our SEO ranking, traffic growth and conversion rates. Our team owns core platform components such as React Server and Node cluster that serve the www.redfin.com customer facing website. ▪ Our team shipped massive tour checkout performance improvement that PMs expect to produce millions of dollars of incremental revenue annually from the initial A/B test. We had to deal with complex business logic and various technical challenges but in the end, we rolled out this win to everyone.▪ I identified HTTP/2 Push as a major feature that can improve our secondary assets (CSS, JS, images etc) delivery. This cross-team effort is still underway but initial numbers show reduced script load time by 75% from 200ms to 50ms across all browsers and devices. ▪To prevent performance regressions, Redfin runs performance tests as part of the continuous integration pipeline that utilizes Google’s Web Page Test, PageSpeed and Lighthouse. But tests had 2 major issues when I joined the team: speed and variability of perf numbers which caused bogus failures. I led the team of 4 engineers to put the new performance testing framework into place which solved both problems. This reduced our dev on-call duties by 90%.▪ Once a month, teams report to the executives how their pages rank in performance against major competitors. This analysis was done manually and was causing hardship. I put daily tests in place and created dashboards that automated the entire flow. ▪We work on improving API performance. This often involves optimizing SQL queries and Java code, various caching strategies and parallelizing execution by utilizing Node.js async programming model.▪We shipped load and performance API tests using Apache JMeter and Yourkit. They allow developers to easily test and analyze how their back-end changes affect performance.

▪ Led multi-team, multi-company effort to integrate showingtime.com automated home booking services for Redfin tours. Previously, booking home tours with listing agents was a manual, time consuming process. I worked with ShowingTime on defining the necessary APIs; PMs, designers, real estate operations and engineers at Redfin to launch “Home Auto Booking”. I created technical specs, schedules, distributed the work and was fully responsible for timely and smooth launch of the new service across various markets in the US. The integration took 6 months and was rolled out to markets on schedule.▪ Worked on various parts of the large system that automatically books tours submitted through the website. The system finds the best touring agent based on calendar availability, business, drive time to and from other appointments, pricing etc. I worked on drive time integration using traffic predictions from Google and Walkscore, various performance tweaks and automated SMSes utilizing Kafka and Twilio. Automated texting is used to schedule home tours with listing agents, send tour reminders to customers and agents, confirm customer identity and tour attendance. “Home Auto Booking” tech stack includes Java, Node.js, React, Kafka, Samza, PostgreSQL and Jenkins.▪ Integrated Confluence Schema Registry for our Kafka pipeline. Redfin has a large Kafka cluster and is using Avro as the message serialization system. I added required Avro encoder/decoder, deployed the schema registry in our AWS cluster, created training materials and implemented monitoring and alerting for the 3 node schema registry cluster.

SnapSpa.com brought spa services to your home or office. Performed market research, designed the website, worked with mobile beauty providers to create the online marketplace and found early adopters. Business failed due to personal complications.

Worked on Silk browser’s core engine and the webview API (C++) which was based on Chromium and WebKit. Worked on bugs and features in various browser areas: from cookies and downloading to rendering.

▪ Worked on a team of three engineers that owned full stack features such as: home searches, map and home details pages. We re-architected the most visited page on redfin.com, the home details page, to decrease load time, improve user experience and SEO. I redesigned the client-side data model based on Backbone to allow for modular, MVC front-end patterns. Backend tech stack was Java, PostgreSQL, Hibernate, GIS and Spring. ▪ Implemented the offer insights feature. Highlight of the work was in optimizing and caching DB queries and tweaking the geographical search algorithm to yield maximum performance while giving most relevant offer insights for a particular home. http://techcrunch.com/2013/02/07/redfin-launches-offer-insights-to-show-you-real-time-data-on-real-estate-bidding-wars/

▪ Implemented clipboard support in WebKit for the WebOS platform. Clipboard allowed users to cut/copy/paste selected text and provided draggable markers used to adjust the selection. ▪ Worked in a team of two to implement asynchronous spell-checking support for the WebOS platform. We used WebKit WebCore document markers, editor and selection controller to autocorrect, revert and spell check the text. ▪ Implemented WebKit scaling support for the WebOS platform which included various optimizations for image loading and scaling▪ Added support for HTML5 features for the WebOS port of WebKit such as application cache, downloadable fonts, local and session storage and geolocation.▪ Worked on bugs and features all across WebCore and WebOS platform port including image decoders, disk cache, resource cache, cookies, security, graphics layer, fonts, loader, editor, selection, HTML5 support, NPAPI plugins and event handling.▪ Added support for animated PNGs to WebCore PNG decoder.

▪ Implemented app store for the WebOS platform: payment, background download service and authentication in Javascript. ▪ Implemented Amazon store application in Javascript using Amazon partner API. Created Amazon service in C++ to handle secure transactions with Amazon when purchasing songs using OpenSSL and libcurl.▪ Implemented native and Javascript wrappers required to support Google Maps on WebOS. Developed NPAPI plugin that allowed Google Maps to access location services. ▪ Implemented Youtube application for WebOS in Javascript using Youtube partner API. Application supported video searching, history and video playback. Communication with Youtube was done via AJAX. History and other data was saved in an HTML5 database.▪ Implemented photo-upload service for Photobucket and Facebook in Java.▪ Developed GPS navigation application in C++ and Javascript for the Linux platform.▪ Developed multi-threaded Facebook application for Palm OS in C++

▪ Developed custom applications in C#, VB using .NET Compact Framework for Pocket PCs▪ Applications utilize bar-code scanners, infrared ports, MS SQL Server CE, networking and XML.▪ Designed friendly and practical graphical user interface.

▪ Co-taught a training course for high-school teachers. Topics included programming in J#, VB and ASP .NET. The course work began with the programming basics and progressed to GUI development, database applications and web services.▪ Provided technical assistance and helped attendees to understand and complete lab exercises.

▪ Designed and developed a web based content management system in PERL on Unix platform to support data exchange between six colleges (more than 400 active users). “Exquisite corpse” application provided an interactive virtual classroom managing assignments, uploads, grading, comments, tracking and gallery.▪ Designed the underlying database schema (MySQL). Added support for uploading art files.▪ Designed and developed an authentication and authorization service using Apache.