About the blog

This blog is nothing more (or less!) than my long tweet threads consolidated into blog entries. Why am I not just writing stand-alone blog posts? Simply, I like the Twitter medium. I like the constraints of having to break an essay down into 280-character chunks, where each chunk has to move the story along. I like that people can respond to discrete parts of the essay, or forward them on with their own thoughts, etc. I like that it lives.

But I also realize that for some people, digesting a long tweet thread is annoying. It’s also hard to bookmark tweet threads, or reference them in other media, etc. So I’ve made this blog. It also means that I have blocked the major tweet-unrolling apps. I feel like if I take the time to build a long-form version of my thoughts, I want people to read the version I built.

There aren’t going to be any comments turned on here - if you want to comment on a post, reply on Twitter!

About me

I’m Andrew Certain. Until June, 2, 2022, I was a VP and Distinguished Engineer in Amazon Web Services. I’m now an independent consult. See my services page for more info.

I joined Amazon.com in March, 1998, and originally worked in the Website Metrics team, building tools to help content owners understand how customers were navigating the web site. From then until the fall of 2005, I held a variety of roles:

I was an engineer in and manager of the performance engineering team, when we developed PMET, the first version of the tool that Amazon uses internally for performance monitoring, and which inspired CloudWatch.

I was the manager of two parts (at different times) of the ordering pipeline. The first time I managed the engines that compute various adjustments to the order total price (e.g. promotions, tax, and shipping charges). The second time I managed the workflows that move orders through the system into our fulfillment network.

I was the Principal Engineer driving Amazon.com’s move away from a monolithic application to a service-based application that could manage e-commerce for other companies.

And I was the first manager of the Fulfillment By Amazon (FBA), starting when it was just me and a product manager working with Jeff Bezos and other executives to define the product.

In addition, I was the head of Amazon’s bar-raiser program (at a time the company was small enough that that role was a part-time, volunteer job in addition to my real jobs listed above).

I left Amazon in the fall of 2005 to spend a year at home with my children, and I then spent about 9 months at a start-up before returning to join EC2, where I was the lead engineer on the Elastic Block Storage, from October, 2007, through March, 2012.

I then moved to our Networking team to build the first release of the software that runs our internal network devices (e.g. the routers that handle all the traffic in our network). In May, 2014, I moved to Database Services to work on an internal database project that was the inspiration for QLDB. I started focusing on DynamoDB near the end of 2015 through the end of 2017, before returning to focus on QLDB.

For the last two years I was there, I was working on making our purpose-built database engines easier to use together.

Other than the success of EBS, the thing I’m most proud of from my time at Amazon so far is instilling a culture of caring about the latency of our systems.

Before Amazon, I worked mostly in the field of computer graphics. After graduating from the University of North Carolina, I wrote rendering software on a variety of platforms (including, in 1993, a home virtual-reality video game platform with head-mounted display that never saw the light of day, as well as a terrain rendering engine for PC video games. I had a brief (one quarter) graduate career at UW working on surface modeling with subdivision surfaces and wavelets, and then co-founded a startup where we extracted tailor measurements from body-scan data to help with mass customization of clothing.