kulikalov

While I recognize the efforts in developing natural language to SQL translation systems, I remain skeptical. The core of my concern lies in the inherent nature of natural language and these models, which are approximative and lack precision. SQL databases, on the other hand, are built to handle precise, accurate information in most cases. Introducing an approximative layer, such as a language model, into a system that relies on precision could potentially create more problems than it solves, leading me to question the productivity of these endeavors in effectively addressing real-world needs.

A simple ngram index can address your concern

Isn’t the end of copyright also the end of IP? Then what?

I actually actively avoid acquiring context around this music, do you think I’m missing out on something?

Jon Hopkins: https://youtu.be/sVefn-AXoZw?si=_AJMSb-vh0bMiyax

I don’t know anything about the author or his target audience, but the music he creates is the most serene and captivating thing I've ever heard.

Put on headphones, lay down, close your eyes and just dive into it.

Subway is filthy and unsafe. No one I know would use it late at night. Most of the worthy cultural events are happening around Central Park and downtown. So, if you want the good parts of New York experience without a slight risk of being stabbed by a crazy person, you better live in a walking distance to one of these locations or commute on a car all the time.

I've been living in Europe for many years. Never needed a car, and moved around cities on a bicycle, public transportation, and sometimes a cab. On a train/plane for longer distances. Safe, healthy, pleasant, loved it.

Moved to the US last fall. One can't live here without a car anywhere outside NYC (and maybe Boston/Chicago?).

The reason is not just the car-centric cities design. It's the individualism culture, the severe income inequality, and the crime levels. I'm not advocating for or against the US culture to be clear, just pointing out the slightly less obvious causes of the car use.

If it quacks like a duck - it must be a duck. Not really.

> In every profession there’s a point where further improvements in skills are so subtle that the difference between one level and another can only be appreciated by another expert

Goal defines the benchmark for proficiency.

Better tennis player scores. Better software engineer writes more space/time efficient code or solves business problem faster (whichever the objective). Better writer gets more views or inbound sales or whatever the goal is.

If you get better at achieving ever greater goals, then your expertise is growing. No amount of luck can compensate for a lack of expertise.

Learning remotely is faster? I thought it’s the same as remote plus extra experience from having work related conversations over the lunchtime

For the engineers who have many years of experience working both remote and on site - how do you compare the pace of learning from your peers and forming meaningful bonds?

How do you form your peer group remotely?

That might be unrealistic in some cases but one way to deal with this is to take control over the situation. If you have a recipe to fix a particular problem whether it’s about unrealistic expectations or people not feeling safe - propose it, take full responsibility for the implementation and push it through. It takes a long while though.

Check out the chapter on embeddings in this wonderful book: https://web.stanford.edu/~jurafsky/slp3/

What are the storage and latency requirements?

It seems possible to create separate tiny devices for different applications and sell them prepackaged. The installation process would look like plugging it into electrical outlet and connecting to your local Wi-Fi. Redundancy can be achieved using close friends and family devices of the same class.

Technically, one can say the same about any devops services out there, for instance. Makes more sense to opensource it.

Still, would you mind sharing the use case and the outcome? I have a couple of clients interested in similar solutions, but so far the potential outcome of this approach doesn't look promising.

I’ve built a similar solution around BigQuery. I used temporary tables to let user iterate over the data instead of writing huge analytics queries.

I would open source it if I had decent experience with OSS.

If anyone with OSS maintainer experience wants to help with that - ping me.

You don’t need ci as a solo developer. A few bash scripts and a docker container will suffice.

Define “deep understanding” here? You certainly have to lean python, at least because you are gonna need it for data manipulation and cleaning no matter what you do in this field.

Not a recruiter, but I imagine there are at least 3 reasons for your repo to be ignored or deprioritized in favor of typical interview: - not everyone contributes to OSS - it’s easy to fake - it’s hard to assess within a short period of time

This it seems to be hard to standardize and streamline candidates validation based on their oss repos.

Also, afaik the goal of interview is to assess candidates thinking patterns, it’s hard to do quickly based on an arbitrary piece of code.

IMO it makes more sense to simply mention the most important skills inside the position description bullet points.

System Design Interview Volume 2 contains a great overview of all kinds of architectures if you are into this kind of things. From there just pick the techs that you don't know and learn. This book is like a table of contents of other techs

Is it unusual? My first year here

How about Public Money, Public Data? For instance, there’s a whole industry around finding public legal information about companies. State’s websites offer throttled database access, some offer data for free, but most selling it for dozens of thousands of dollars (officially)

Reminds me of Julian Jaynes theory

Could you describe a scenario where you need to resume a conversation?

Isn’t GPT a trademark owned by OpenAI? Is it legal to use it?

The world have moved beyond zero sum economy a long time ago.

What you're describing requires AI to make conjectures, hypothesize, design experiments, run them, and assess their results.

This is great! Thanks for BigQuery integration, I was passively looking for an alternative interface. Looking forward to try it. Possible feature request (not sure if you have it) - BigQuery creates temporary tables after each query run. When exploring the data it would be super cool to write queries sequentially by using those temporary tables. I do it in BigQuery UI, but it's quite tedious.