Amazon announces Q, an AI chatbot for businesses
cnbc.com
cnbc.com
Amazon's AI team seems to be the B-players who couldn't find a job at Google, Apple, OpenAI, Facebook, and friends, and things they've delivered were always less than impressive. Toxic work environment doesn't draw the best people.
I use a lot of Amazon services, but their code completion was abysmal compared to co-pilot. Whisper wipes the floor with Amazon Transcribe. Etc.
The pricing is also questionable, at $20-$25/user/month. I use generative AI a lot, and my OpenAI API bill is a fraction of that.
It's also not well-integrated with the rest of AWS, where I want APIs rather than per-user services.
The upside is less than clear.
To be fair, nothing is well-integrated with the rest of AWS
Google, Apple, Facebook, Amazon, Netflix, every other tech company
Companies that are leading in AI:
OpenAI
I have similar thoughts in response to a lot of his comments. His criticism is often precise and fair, but it often comes along with an insistence that his "world model" is the one true solution. Maybe that is true? I haven't seen the evidence of it as of yet.
There are no great products in the pipeline for OpenAI. Microsoft is doing the productization.
But if you are even not the best of the best, but the worse of the best, why the hell would you ever want to work for Amazon?
Source: former AWS employee (Professional Services). I’m not bitter, it was my 8th out of 9 jobs in almost 30 years of working. I made my money. Learned what I could. Put it on my resume. Got Amazoned with a nice severance package and found a much less stressful job doing the same thing in less than 3 weeks
Facebook and Apple offer the same $$$, with much less toxicity. If you can't get a job there and need money, Amazon is a rational choice. At my cost-of-living, if I worked for Amazon for about five years, I could retire.
In my current job, I get half to a quarter of what I'd make at Amazon, but enjoy life. That's worth it. But if it stops being meaningful and fun, perhaps at some point I might go somewhere where I can pay off my mortgage in two years.
I specialize in enterprise app dev + AWS. The logical place for me to at least have on my resume is AWS Professional Services. It was the quickest , easiest method to get into BigTech.
There was no coding interview, it was fully remote and there was a never a return to office mandate.
One fortunate thing about AWS is they have a very easy approval process for ProServe consultants to sanitize their work and publish it to the open source AWS organization on GitHub and its MIT licensed.
https://github.com/aws-samples
I open sourced everything and forked it to my private profile. It gave me a nice open source portfolio that I could say was used in real world projects and I was able to use 5 of the 8 projects at my current job. Two of the others were using now deprecated techniques.
The pay increase wasn’t that dramatic. It was about a 35% pay difference than what I was being offered pre-Covid.
It was about a 20% pay cut to leave and work for a smaller consulting company full time.
You would absolutely hate yourself trying to survive at AWS for five years. I knew after the first year I wasn’t going to be able to take more than my initial four year offer and I planned according.
https://news.ycombinator.com/item?id=37969302
As far as enjoying my life, I would never satisfy my current work life balance for working at any large company.
I don't think you know my current job or my level of seniority.
1) I am a sufficiently well-known developer that, while you wouldn't recognize my name, I am a major draw when I give presentations by name recognition in my field, and people have asked me for e.g. autographs before. You would have heard of the project I am best known for, even if not who built it.
2) I am not a "standard CRUD developer working in Atlanta." I work for a non-profit. It's engaging, meaningful, and impactful work, and it pays my bills.
My last job offer from tech was at double my current level of compensation, and that wasn't a FAANG, but a late-stage startup. I know peer salaries at FAANG, and a quarter is a pretty good estimate. I know people at FAANG making 10x, but I'd need to get very lucky to get that (a job lining up with my unique background). It might be 2x if my background completely wasn't taken into account (which it wouldn't be, for many roles).
There was an “anonymous” internal Slack channel called #pay-equity. Wheee you could use post your compensation via A workflow (?)
You make a lot of assumptions about whom your talking to.
Life is short. If you are spending half of your waking hours in a job, it might as well be fun, meaningful, and helping the world.
If it pays the bills, and it's what you'd be doing if you were independently wealthy anyways, why not skip the "independently wealthy" step if someone is glad to pay you to do it? That saves a few years of your life. I am doing _exactly_ what I'd be doing with my life if I had a few million dollars in the bank and could spend my life however I liked. Indeed, having a sponsor opens doors and provides support for it. And the BATNA, if my employer were to ask me to do something I don't want to do, is to become independently wealthy at a FAANG.
As a footnote, seven-figure salaries at FAANG are common for world-class experts in many fields (not just in AI, although that's where they're the most common).
There used to be a cartoon:
https://en.wikipedia.org/wiki/On_the_Internet%2C_nobody_know...
You can take what I say at face value, or you can decide I'm a liar. It's up to you. I'm checking out.
The canonical example one was that the creator and maintainer of homebrew couldn’t get a job at Google even though most of the developers used his code everyday.
More than likely as an L5 you’re going to get an offer around the mid 200s at an L6 you’re going to get around a 350K.
And it doesn’t matter how good of a developer you are if you haven’t done or practiced a coding interview or system design interview. You won’t get to an L6 unless you show that you have worked and shown “scope” and “impact”.
Apple is similar.
Facebook has FAIR, which is pretty world-leading, and OPT-175B is not bad. No connection to product, but the core research team does well.
Netflix has no real AI, although they were in a leadership position about 15 years ago.
Amazon kinda stinks. They have a lot of mediocre researchers. Alexa works pretty well, but by most accounts, consists of a bunch of hardcoded algorithms. Just throw people at the problem. The rest of the stuff is unimpressive as well.
Which just had a major update.
https://aws.amazon.com/blogs/machine-learning/amazon-transcr...
I assume this is named based on Q from Star Trek.
More discussion over here: https://news.ycombinator.com/item?id=38448137