Spotlight: Anne Aaron, Director of Video Algorithms at Netflix
A couple months back, Business Insider feature its current list of Most Powerful Women Engineers of 2017. Among them is Anne Aaron, an engineer from Netflix. The story was also picked up by local media.
The name sounded familiar but it was during my trip to Barcelona during Mobile World Congress that I heard about it again. This was during the Netflix house session with Todd Yellin. One of the senior engineers (a quantum physicist) was telling me that his manager is a Filipina and that I should meet her.
A couple of weeks later, I was headed to California for a familiarization tour of the Neflix HQ. As I was checking in my trip on FB just before my flight, schoolmates from college left comments telling me I should met up with Anne as she works for Netflix. I said I will have a one-on-one talk with her about her role in Netflix. Only then I was told that she was actually a fellow alumna from Ateneo. We were in a parallel course too — she took a double major in Physics and Computer Engineering while I took Chemistry and Computer Engineering. She was just one batch ahead of me and a summa cum laude graduate as well.
Anne took her masteral degree in Electrical Engineering at the Stanford University and also finished a PhD in the same university.
Anne has been with Netflix for over 5 years (she joined the company in 2011). She started off as senior software engineer and worked her way up to become Director for Video Algorithm at Netflix.
Anne’s work at Netflix has been heavily focused on delivering the best quality streaming at the lowest bit rate possible. One of the many goals is to give the best user experience to the customer by providing an optimal condition to stream the entire Netflix library.
And they’ve made a lot of strides in approaching the problem on many fronts. Among them is the complicated world of encoding. At first, they tried per-title level encoding, a pioneering strategy in video streaming. It was good but some titles turned out better than others or some scenes looked better than the rest for the same title. Today, they’ve gone to scene-level video encoding where the Netflix AI tweaks each individual scene and figures out the right encoding to give it the optimal quality for streaming.
On the hardware side, Netflix deploys their own Open Connect Appliance which basically caches the titles at the ISP-level or Exchange-level to deliver the content faster to subscribers. These servers are given to the ISPs for free.
Anne says the work is fulfilling as all the efforts allow Netflix to reach more customers especially in regions where connectivity isn’t that optimal. Soon, Netflix will be rolling out video algorithms that will deliver DVD-like picture quality at just 100kbps bitrate (excluding 56kbps audio). That’s already an impressive achievement there.
Before heading to Netflix, Anne spent a little over two years at Cisco working on Flip Video. She was doing engineering lead for the video decoding and video encoding back then until Cisco killed the Flip Video project in 2011. New doors opened up for Anne and she got invited to join Netflix for the second time (the first time was 2 years earlier but she skipped it).
Today, Anne heads a sizable team of researchers, scientists and engineers at Netflix, helping build the infrastructure and technology behind the streaming service.