Working with Instagram Reels systems to quickly and reliably deliver Reels content to the end user.
I joined the Reels Recommendation Foundation at Instagram shortly after graduating college. I found myself thrust into a world where millions of users are a drop in the bucket, and downtime means serious consequences. I had to learn a lot about reliability, and how to make systems work with such monumental amounts of data being gathered and generated.
I built a number of systems used in Reels, including a pipeline used to generate topics for Reels using video and audio machine learning models. I was also able to work on offline ranking of Reels, which ranks pages of Reels even when the user is not in app, at off-peak compuational times. This helped save computational resources when they were most needed, at peak.
It was fascinating seeing how a company with nearly 100,000 people runs on a daily basis. While a lot of the processes were fairly inefficient, it was amazing just seeing a company of that size align on direction at all. Although I ultimately left Instagram to pursue General Task, it was an amazing company, and I was able to learn from some of the smartest people in the industry.