> Design is more fun than making it work.
Great wisdom for any kind of project
I’ve implemented LLM based “intent processing” instead of old school “give it a list of every possible phrase that someone could use to activate an intent” for call centers for a couple of states and a couple colleges - I worked for AWS ProServe directly in the pub sec division and now work for a third party consulting company.
Think of an intent as something like asking Siri for driving directions, setting an alarm etc.
One of my specialties is Amazon Connect for hosted call centers - based on the same service that Amazon uses internally. Think of the difference between how Google Assistant and Alexa worked pre LLM and how Siri works today compared to tool calling that modern LLMs used.
That’s just using an LLM in a product.
As far as how I use it everyday? I am a staff consultant and in the previous times, I needed at least a couple of developers under me just to get the grunt work done on time. Now I can just treat Claude Code and Codex as faster, more accurate ticker takers.
Of course I’m not going to dox myself (more than I probably already have) or break any NDAs
And before the gate keeping starts about “I must be inexperienced”. My first time coding was on a 1Mhz Apple //e in assembly and BASIC in 1986. During the first decade and a half of my professional career, I programmed in C targeting everything from mainframes to PCs to Windows CE ruggedized devices.
> How to Have a Bad Career in Research/Academia Pre-PhD and Post-PhD (& How to Give a Bad Talk)
I've also not gone through the whole presentation, but does he at any point talk about the moral choices one will most definitely have to make during a career in tech? (this is related to the previous paragraph). Is it a "bad career" if people choose not to work for companies (such as Alphabet) that have gone all in behind AI? Seeing as now AI is used by State-entities for very nefarious reasons. Like I said, 2026 is way different compared to 2016.
The challenge for computer science researchers who have qualms about working for Big Tech is finding an alternative career path. Speaking from an American point of view, academia has always been competitive, and the immediate future of research funding is uncertain given the political climate. This uncertainty also extends to government labs. The challenge with industry research is that there are not a lot of non-Big Tech employers of computer science researchers. This leaves starting a business, but business is very different from research.
I’m a tenure-track professor at a community college in the Bay Area. While I’ll never be able to afford to purchase a home near my job, I am able to live well as a single man renting an apartment. I have a great career teaching and using my long summer breaks for research and side projects. I like not having to worry about “publish or perish,” and I enjoy teaching and mentoring students. While this might not be considered “successful” for some people who are aiming for a professorship at an R1 university or an industry job at a top company’s top lab, I love my job and believe it’s a fantastic route for someone who enjoys teaching and who also wants extended time during the summer for research and side projects.
In aggregate, sure, but no company today comes within an order of magnitude of the power an IBM of the ‘70s and ‘80s or a Microsoft of the ‘90s and ‘00s had over the tech landscape.
Big Tech today is the media. Taken together, they completely control what a majority of the populace knows about the world. It is considered completely impractical and somewhat suspicious not to carry one of their location tracking communication devices at all times.
Orwell's Ministry of Truth could not dream of what Meta, Alphabet, OpenAI and Apple can do at any time, anywhere.
In general, Academia lacks sufficient resources and appropriate structures for dedicated efforts.
Pay is abysmal and politics is toxic. Both scare away lots of technical talent.
That being said, it's certainly different for researchers. I can imagine that being a researcher at Google is more fun than being a median SWE in another FAANG. But still, I find these companies less enticing in general, even the products tend to degrade as they keep pushing the monetization.
I think it depends on the interests of the researcher. If a researcher is comfortable being a “brain for hire” who is comfortable solving research problems that are driven by business needs and where there needs to be short-term or medium-term results, then I think there are plenty of opportunities at large companies, including the FAANGs. I find research more fun than software engineering, but researchers are far from immune from pressures to ship.
If a researcher is more interested in curiosity-driven work and who wants to work on a longer time frame, I’m afraid that there’s no place in industry, except for maybe Microsoft Research (which I’ve heard changed under Satya Nadella), that supports such work. The days of Bob Taylor-era Xerox PARC and Unix-era Bell Labs ended many decades ago, and while there were still curiosity-driven labs in industry well into the 2010s, I have witnessed the remainder of these old-style labs change their missions to become much more focused on immediate and near-immediate business needs.
One of them (Patterson & Hennessy) was, and the other was (is?) chairman of the board, I forget which.
Nowadays? I wouldn't touch anything that comes from Google (granted also not from any other big tech company) with pliers.