My Amazon Experience

I wanted to share my experience, it has ben some time I left Amazon but never shared this on public Blind channel. Many of the things I will say may identify me, but I give a damn.

I joined Amazon as L5 Research Scientist, Bay Area in the Alexa team working on screen devices. I wanted to work on core Machine Learning problems as this happens to be my background and interest. After joining I was told that I will need to work on designing experimentations, which was ok as it is something I learned, but never wanted to do it long term.

Then reality struk, the idiot PMs idea was to tweak small changes in the display to increase engagement by a big margin. Kept uttering like an idiot “if we don’t do this, we miss our A-team goal”. I thought I would work on personalization and ranking and not spend my time generating json files that determine how cards will be displayed. It was a complete waste of my time. I knew the team can’t offer me anything, and frankly they didn’t need to hire me or anyone, they could have used a scientist from another team for this job. I tried to get into a more ML focused team at A9. Most positions at A9 were Applied Scientist positions and not Research Scientist, which made team change difficult, especially give in 4 months I have done mostly stupid work. I went through 3 tech interviews, unfortunately it wasn’t upto the mark and didn’t work out. At that point I decided I will leetcode and get the hell out after my 1 yr vest.

During this time the director of the group left and I had a new manager but kept working on the stupid project till December 2019 with idiot PMs. I didn’t mind and kept leetcoding. Then when I finally started in the new group in early 2020, there wasn’t any good project and that manager left for Alexa Auto. I was assigned a new manager who was based in Seattle. I didn’t care as I was set to leave and was doing bare minimum work. My new manager (who was my last manager) is a good person, unlike most managers I hear about on Blind. I did get to do some decent work on experimentation under him but not the kind of ML work I wanted to.

Then came March 2020, COVID struk and almost all companies went on hiring freeze and daycares closed. It was very difficult to adjust given the new situation and to interview with a toddler at home. I kept slacking at work, and got put on the dev list. I will say that I don’t blame my manager for it, it is something justified. He did help me with clear guidance on deliverables and fully supported me. Thanks to him, I came out of the dev list in late August. I kept doing an ok job after that and set my goal to change after my year 2 vest. In 2021 my forte feedback was “needs improvement” but I didn’t care, as I was concentrating more on leetcode and interviewing. Finally I did get a few offers and accepted one. During exit my manager informed me that I am not eligible for rehire at Amazon and I told him that not to worry as I won’t be considering Amazon in the future. Although I got a huge TC bump, I didn't leave for TC, I am looking for interesting work and good culture.

TC while joining Amazon: 292k

TC while I left Amazon: 315k due to stock

TC while joining Roku: > 510k

Current TC: > 550k with recent base increase and additional stock grants

#amazon #pip

Edit1: I work with excellent PMs at Roku and respect them. The PMs I worked with at Amazon are complete idiots.

Edit 2: Addressing the comment if this is related to technically incompetent managers. My experience says no.

Managers:

1. My 1st manager was an L7 SDM. He was very supportive but not a science manager. I feel he sensed I was not happy with the project. He was open to me spending part of my time with the other L7 science manager who played a crucial role in my hiring and who was my 2nd manager after the re-org. However, it was only for 3 months he was my manager and then the re-org happened.

2. My 2nd manager was very technically competent, PhD with years of experience. But he was a bully, a micro-manager and the 1-1 felt like harassment. 

3. My 3rd manager was a science manager, technically competent and a decent person.

PMs: All of them were terrible.

1. Already told about the 1st PM, not going to repeat it.

2. Another PM wanted to make a small tweak to the Alexa feature, run A/B tests and show improvement. Improvement can also mean decrease in -ve engagement. Given the number of metric and scale, one can easily show a statistically significant improvement for a reasonable tweak. I was tasked with running sql scripts and writing reports. Great for the PM's promo but terrible for my growth.

3. A PM said she will not let me start an A/B test unless I include more features for my model and do an evaluation. Even though I showed significant improvement in off-line metrics with the features I used. And this after literally fighting to get time from her and she gives me this input just before jumping to another team!