Oct 2021 – Oct 2022 · 1 yr · San Francisco Bay Area

Oct 2018 – Oct 2021 · 3 yrs · San Francisco Bay Area

• I lead multiple teams on Assistant Infrastructure. We work on the core infrastructure components that enables our product team to build quickly for new use cases for the Assistant and scale easily to provide our users a natural and reliable assistant experience.
• Collectively, my teams are responsible for:
• The core Assistant serving platform, its reliability and long-term architectural design and evolution
• Managing the state and context of the conversation, in the presence of concurrent requests and multiple devices
• Moving the assistant server platform to micro-services to improve maintainability and developer experience on the platform
• Scaling Assistant features to work seamlessly across all devices by providing the right abstractions for feature developers while allowing for device specific tweaks
• Execution and dispatch of the end user task (mutating the state of the world) across various execution targets
• The core communication protocol that Assistant devices and the server stack uses to communicate the user request, response and actions

Dec 2017 – Oct 2018 · 10 mos

• I moved to Bangalore, India to help grow the Uber Bangalore site where I lead the Vehicle Solutions team. We build the driver-partner on boarding flows and third party integrations (rentals and financier companies) that result in a vehicles marketplace for drivers to choose from when they want to drive on the Uber platform but don’t have or can’t afford to buy a vehicle. We built a Rentals Platform API to scale this worldwide and expand our marketplace choices for our driver partners.
• I also bootstrapped a new AdTech data and measurement charter out of Bangalore. This is AdTech on the buy side - where Uber is the advertiser and we’re working on optimizing our marketing spend and improving our acquisition and engagement ROI.

Mar 2016 – Nov 2017 · 1 yr 8 mos

• I grew and led the Marketplace Health team whose mission is to ensure Uber’s markets are healthy and operating at maximum network efficiency. This involved coming up with independent ways to evaluate the health of the marketplace (reliability and efficiency metrics), aligning the tech and ops orgs around those metrics and helping them improve health in their cities.
• The team built metrics pipelines and data visualization tools for analysis, complicated data science models and execution at scale to simulate flow of cars to identify maximum efficiency, and alerting algorithms and tools for ops teams when health was critical.

Sep 2017 – Oct 2022 · 5 yrs 1 mo · Mumbai Area, India

Jun 2013 – Mar 2016 · 2 yrs 9 mos · San Francisco Bay Area

• Started as an Engineering Manager by bootstrapping the TV Ads and Analytics teams. Over the ~3 years, we grew to a 27+ group of engineers and PMs.
• TV Ads:
• We build one of the first realtime TV ad insertion systems at the set-top box level, so every user in theory can be shown a different ad depending on their profile and interest.
• Ventured into other potential ad offerings for non-live content
• Analytics:
• We built the analytics for all of TV and Internet products
• We built ML models for understanding customer interests, churn risk, etc.

Jun 2007 – May 2013 · 5 yrs 11 mos · San Francisco Bay Area

• We built containers before Docker and Mesos were a thing!
• I worked on the Linux Kernel team for our servers, and among other things built:
• 1. A prioritized IO scheduler before CFQ became the default Linux IO scheduler.
• 2. A proportional IO scheduling mechanism that is now part of the CFQ IO scheduler (works with cgroups)
• 3. Memory containers - including anonymous, pagecache, kernel, slab, and more as part of the memory cgroup accounting mechanisms. Rolled out memcg @ Google!
• 4. Testing harnesses for our large scale applications (before we roll out a new kernel there)
• I also worked on the memory management layer for Google's cluster management solution. We completely revamped how Google binaries share memory resources on our servers!

Aug 2006 – Dec 2006 · 4 mos · San Francisco Bay Area

• Intern with the CrawlIndex SRE team working on extending Linux Kernel to support a prioritized IO scheduler. This work then made its way to the production kernel team at Google, as I joined them fulltime in 2007.