Meta

At Meta I’ve worked on 3 different teams: XROS, Facebook Newsfeed Ads, and Meta Rayban Glasses.

XROS

On the XROS team, I served as team lead for a team of 5 building UI components and widgets like text, scrollviews, and the application model for the XROS operating system project. The operating system was novel and not based on unix/linux. The work was much more similar to working on distributed systems than working on traditional embedded systems, although we also faced unique hardware constraints particularly in our graphics rendering stack. This work gave me an opportunity to dive deep into how operating systems work, developing new programming languages, how UI systems work, graphics rendering stacks, working with applications distributed across several processes, and where there is room to innovate and improve on the operating systems and application models of today especially as we eye potential for new application models in AR and VR including declarative UI and server-driven UI vs traditional UI kits.

Facebook Newsfeed Ads

Facebook never showed ads in the top position of news feed for a few reasons. It was thought it would be a bad user experience to open the app and immediately see an ad at the top of feed, and the time required to prepare an ad for the top position was significantly longer than the time to prepare an organic story from a friend. When I joined Newsfeed Ads in 2021, I was tasked with delivering ads at the top position to unlock an estimated multi-hundred million dollar opportunity.

When making changes to a high traffic area like newsfeed, it’s critical to ensure the tradeoffs are worth it. Any delays in feed loading can trickle down to impact DAUs and tons of other critical metrics. Depending on the financial situation of the company, the tradeoffs around timespent vs ad dollars fluctuates, but its always better and easier to ship with the best tradeoffs (max revenue and min impact on timespent). Any action taken needs to be thoroughly tested and the tradeoff proven before shipping.

My joining the team coincided with the start of a project to change how ads were delivered which had a side effect of making it faster to receive the first ad. However, this still required significant tuning as the less time spent preparing an ad directly impacts the goodness of that ad in terms of relevance to the user. In addition, we encountered several issues on the Android and iOS app side around how changing expectations for first ad position impacted existing delivery systems and application startup.

I led discovery on this project and started leading a team of 3 in 2021, which grew to a team of 10 in early 2022 and then to a team of 40+ in late 2022 as we neared shipping. I left the project in late 2022 to join the Rayban Meta team, and the project ended up shipping in H1 2023.

Rayban Meta Operational Excellence

I joined the Rayban Meta team in late 2022 right after the Rayban Stories had just shipped. Work was already underway for the next generation named Rayban Meta. I ended up on the operational excellence team, focused on defining and tracking key operational metrics for the engineering team and the project as a whole. My work consisted of building dashboards, defining metrics, collaborating with partner teams on QA, bug triage, and customer excellence, and in late 2023 I spearheaded work on a project to automate parts of code-writing using LLMs.