Can someone design a chat system that actually remembers conversations? I just had this issue with Apple support. If you don't reply immediately you get a new session with a new representative and the new representative doesn't have any of the information from the previous session. Extremely frustrating to have to keep sending the same information again and again. It took a call to finally sort it out.
What people miss is that micro services/SOA is an organizational concept more than anything. At Amazon, SOA is tied to the concept of the two pizza team. Each 2PT completely owns a collection of related services, owns the roadmap for those services, and can make deployments independently of any other team. If your company doesn’t have enough engineers to justify at least say 5-10 scrum teams then you might not be large enough to need microservices.
Firefox keyword searches are the way to go - you can setup URLs with arbitrary %s variables and go directly to the results page for a given tool. I have one or two char shortcuts for my most commonly used tools setup and my productivity exploded. With bookmarklets you can embed JavaScript and do things like implement parameters or loop through lists of search terms with each term opening a new tab. Seriously one of the most underrated tools available for anyone that has to use web tools all day.
This is a big part of Amazon’s success. Service oriented architecture gives teams enough autonomy to define their own processes. “You build it, you own it” often extends to products as well as services.
It also helps that Amazon isn’t afraid to entertain ideas from lower in the ranks. You just need a strong idea and be willing to write a memo that explains it, then work your way up the chain until you get to a decision maker.
I’d say we’re closer to an order of magnitude larger When you consider all the different businesses and domains. My org alone has about 1000 services. But we have massive scale and the employees to support that number of services. You also won’t find one architecture; it runs the gamut from lambda to REST/RPC to pubsub, event bus and workflow architectures. Typically new orgs start out with a few larger services that get broken out over time, as the complexity and number of employees continues to grow.
Amazon does micro services (or SOA) extremely well. In fact they practically invented the concept. It’s intricately linked with the 2 pizza team and service ownership concept (you build it, you support it)
The problem with UBI is that most versions of it would essentially liquidate social services that people would normally receive. UBI only makes sense in addition to the other parts of a comprehensive safety net (universal housing and healthcare), not as a replacement.
I’ve found OneNote very useful for Todo lists. It’s very flexible and supports collapsible nested lists. But the killer feature is the “dock to desktop” feature, which makes the list always visible on my monitor. It forces me to stay on task and reminds me what to do next. I create a new page every week and copy over my remaining tasks from the last week. I’ve been doing this for 3 years now and haven’t looked back.
What happened to the company that had a generous warranty policy? I went through 3 keyboard replacements for free and got a newer model each time. I miss that.
I’ve been at Amazon 10 years and this post sums up a lot about what I like about Amazon as well.
Regarding 2PTs: the big thing here is that each 2PT owns services, and has complete operational and (usually) product responsibility for those services. “You build it, you own it.” There is still org level P&L but at the director/VP level and not at the 2PT level. Also there is only one middle manager level before the executive levels (7), which has a nice dampening effect on bureaucracy.
Finally, the working backwards culture has a lot of innovation opportunity. Many large divisions started with a low level employees vision doc.
Only if the seller pays more to keep their products separate via stickering. Otherwise the seller’s products are commingled with other sellers’ products and it’s impossible to know which seller’s physical inventory was used to fulfill the order.
It’s worth noting that nothing Amazon built internally was ever directly sold through AWS. Rather, Amazon built something similar internally then used its engineer expertise to build a completely new product for AWS. Selling your big ball of mud directly is a recipe for disaster.
To their credit, Amazon is pretty diligent about making sure that high performing employees get pay raises to match their market. My record is a 30% raise, mostly by switching job functions from non-tech to a tech role.
The article overstates the use of ML at least in fulfillment we use a lot of math models (eg xpress solver) and optimization algorithms (traveling salesmen type problems) but not so much in the ML space.
What’s interesting is seeing SOA evolve: first it was service calls everywhere, now it’s increasingly asynchronous publish/subscribe architectures (think AWS SNS and SQS) and service orchestration frameworks (SWF, Step Functions). In my department we realized that 700+ services is untenable and we are now exploring home grown frameworks and SDKs to help manage those levels of complexity.
It’s notable that Amazon, despite record growth, is also in a hiring freeze for nearly the entire Retail business. Leadership is also simplifying the management chain to reduce bureaucracy.
Rather than all-you-can-stream, charge a low rate per song play to users. Assume that existing services charges $15/mo, and heavy users might play 16 hours a day, and act song length is 3 minutes, that comes out to about $0.0015 per play to be the same value proposition. But guessing average is much lower so we could up the rate 2x or 4x and customers come out the same. Everyone gets the same cut of revenue per song play, so popular artists get the same cut of streams as indie artists.
Amazon is so decentralized that the anti-patterns you list are probably only under your VP. If anything I think the company is taking action to trim bloated orgs. For example there are internal efforts to prune middle management issues.
(For those that don’t know, Amazon has only one middle manager level (L7); L6 is a frontline manager and L8 is a director. The anti-pattern is when you have L6 reporting to L6, L7 reporting to L7, L8 reporting to L8, L10 reporting to L10...)
Interestingly Amazon doesn't offer any incentives to get an MBA if you're already working there. No tuition reimbursement, raises or promotions. You're more valued as an external MBA hire.
In WA we have absentee ballots which is really the best of all worlds. You fill it in, sign it and mail it back. Couldn't be easier and you don't have to worry about polling locations. They are also hardcore about checking your signature; If it doesn't match your voter registration the ballot isn't counted and you get a letter in the mail.
Anecdotally I know, I got my foot in the door at Amazon by temping and later being hired. Amazon is a great place to start as a temp IMO because they are always looking for talent and you will often be embedded directly with teams and be treated as a regular employee. I also know several former FC employees that moved HQ.
Similarly, Apple Store employees can work their way up to HQ.
Does anyone have an example of a detail page with an externally hosted product image? I'm like 90% certain amazon hosts its own images. I wonder if the sellers are using external URLs to upload their listings through seller central, and Amazon automatically creates the listing with the placeholder image hosted on Amazon.