Amazon.com - Client side metrics
The client, page owners of Amazon websites (retail, prime video, audible, imdb,
twitch etc.) needed a way to measure the latency impact of their changes, track
the user behaviour when interacting with the page and be able to create
monitors and alarms to guard for regressions in user experience.
Reduce interation failures rate by half from above 3% to 1.5% which
reduces customer fustration and helps in preventing the loss of trust and
money; last known incident was estimated at >$1b loss. Achived this by
providing an API to track interaction failures where the page will not
responde to a client action.
Increased the mobile platform data coverage for latency and
customer behaviour by ~11% which helped teams take better decisions
on featurures that are presented to clients. This was achived by working
with mobile team to identify untracked interactions.
Saved ~$4.5M yearly by reducing the volume of data generated
from above 1172Tb/day to 50Tb/day by working with teams to identify
duplicated and unused metadata and creating an opt in API to generate
data only on client request.
Skills: JavaScript, Java, Python, Ruby, AWS, Data processing & analytics