Peak is Over

The peak is over! I can now officially say that I’ve been with Amazon for one peak. It sounds funny at first but the time spent at Amazon is measured in peaks. “How long have you been with Amazon?” “Seven peaks.” (Wow! How did you survive that?) “Three peeks” (Well done, can you help me with something?) “This is my first peak.” (Oh, just sit, watch and try to stay out of everyone’s way.)

Peak in supply chain begins a few weeks before Black Friday at which point it is mostly inbound, getting as much into the warehouses as possible. It continues throughout Cyber Monday when outbound gets crazy as Amazon tries to ship all the purchased items to customers. Without any recovery time for the network, Christmas are creeping up on everyone which means more inbound as well as more outbound and also a lot of balancing transfers as the warehouses hit 100% fullness. And then the fullness goes higher and higher and it makes you wonder how it actually looks like “out there”.

Peak is very special time indeed. People work around the clock, emails with special metrics are flying around and the network is pushed above and beyond its limits. Every package that is not delivered before 25th could ruin someone’s Christmas and no one wants that, right? And then you look at the latest metrics and see thousands of crying children who will not get their legos and barbies from Santa. And then you find out that some people in the office have not seen their kids for a week or more. What seems most stunning to me is that people do not complain (as I am sure they would in Czech). They actually do the exact opposite; they take a special pride in being part of this craziness. They pull in extreme hours and build strong relationships, sitting together in the office long after their kids have fallen asleep, only to be the first one back in the morning, to be the first one to look at the metrics and find out what needs to be done to keep the network going.

The offices got empty over  Christmas. The only people working between Christmas and New Year seem to be newbies (like myself) who had not acquired any vacation yet. The campus literally turns into a ghost town. I wonder what is going to happen when everyone gets back. I guess we will try to measure how good or bad the peak was and slowly start working towards the next peak which is always expected to be even bigger (in terms of number of shipped items) than the last one. There is also going to be an after holiday party for all employees and their +1s. There are not many places that are big enough in Seattle, so it comes as no surprise that the party will be at the CenturyLink Field stadium which has a capacity of 67,000 people!

Meanwhile, our team (FEDS for Fulfillment Execution Data Science) is growing like crazy. It is a fairly new team formed in September after a bigger reorganization (yeah, looks like those happen everywhere) and it had around 12 people when I interviewed. Now we are close to 30 people and still hiring. Sounds like I may become a senior team member pretty quickly :)

Do you say you’re interested in what I do? I suspect a lot of it is confidential but I can at least mention the technologies in use. So far my work has been a lot of R and also Shiny, which is a wrapper around R for building online dashboards. There used to be a lot of Tableau dashboards around but the licenses turned out to be super expensive so we try to do everything in Shiny now. That being said I have to mention SQL which I write almost as much as I used to when I worked as data analyst. At least we have a Redshift database which is really fast even with billions of rows and terabytes of data. I also spent an awful lot of time trying to understand some code produced last peak by someone who is no longer around. I decided I do not like code written during peak (when people take a lot of shortcuts and not getting a code review is not the worst thing that may happen). I wrote my first machine learning algorithm in Python; working on it was quite exciting but the results were on par with the baseline model predicting simple average. So much for data science…

Comments Are Closed