1,682 karma · joined June 7, 2017
https://www.theatlantic.com/economy/archive/2025/11/mamdani-...
As some of the replies note, it has been rather successful and popular in other cities like Berlin.
So it isn't a big $10K+ or bust argument. But the sad fact of the matter is very few pet owners assume ownership while planning for the potential bad days and the extra financial burden and responsibility that they should feel when they take on pet ownership.
[1] - https://addons.mozilla.org/en-US/firefox/addon/tree-style-ta...
Most of the complaints here ironically are from people using a bunch of tooling in lieu of, or as a replacement for vanilla python venvs and then hitting issues associated with those tools.
We've been using vanilla python venvs across our company for many years now, and in all our CI/CD pipelines and have had zero issues on the venv side of things. And this is while using libraries like numpy, scipy, torch/torchvision, etc.
CoVar is a small R&D company specializing in machine learning and software development for defense, healthcare, and manufacturing applications.
We're looking for a jack-of-all-trades who can: prototype computer vision algorithms, train and validate classical and deep-learning ML models, build Python web services, develop Javascript visualization tools for time-series data, ... any or all, with the only hard requirement being eagerness to learn.
Contact careers@covar.com for more information.
Some examples of phrasing that would have worked just as well:
- Low-cost developers
- Unskilled, inexpensive developers
But that doesn't quite have the ring of some good ol' casual racism/stereotyping that everyone can pile on I suppose.
1. Multi-line text support and auto indentation support. This is *huge*. Most python REPLs are terrible at this including the default interpreter. You can easily copy paste code from scripts/modules into the REPL and the interpreter just handles everything seamlessly. It is even smart enough to remove a global indent across all the pasted code (if you copied code from within a function that was already indented 1 level up). It makes the REPL experience really really smooth.
2. Tab completion works beautifully with hover overlays
3. The integrated variable viewer is extremely good and you can easily view the local state of your interpreter and explore data/variables. The integrated Pandas dataframe and Numpy array viewers (available even in the free version) are really handy as well.
4. You can even attach a debugger to an interactive REPL session and if you then have breakpoints defined in associated libraries in your Pycharm IDE and then invoke code that would hit the breakpoint, it will pause at the breakpoint and give you the full debugging experience. This is really handy for reducing the time to debugging and investigating issues in code.
5. Matplotlib eventloop is handled very well in Pycharm which basically means that interactively plotting in the IPython REPL using matplotlib works seamlessly.
6. You also get some amount of linting/error checking in the REPL and also syntax highlighting, which is really helpful as well.
7. The IPython interpreter *is the default interpreter* which means that even when you debug code (with breakpoints for example), you get all of the benefits above while debugging, which is a really nice experience, especially with having access to the variable viewer.
8. Another annoyance I had with VSCode last time I tried using it is that the debugger while vastly improved still only allowed single line of code entry and was generally clunky if you wanted to paste multiple lines of code into the debugging REPL. Since you get the full IPython shell in Pycharm at all times (debugging or otherwise), it ends up being a lot more powerful and easier to use.
9. This is underrated, but Pycharm actually has a button that displays a log of all your code entries into your REPL. This is really handy in my experience as you can prototype code in the interpreter with working data/state, validate that it works right and then grab it from that window, copy it, and then paste it into a script/module to "graduate" it to more matured code.
That's what I could muster up off the top of my head. Pycharm in general has a ton of other nice things going for it, but ultimately, it is the really smooth REPL experience and how well integrated the shell is with the IDE that makes it my go-to IDE for anything Python.
With VSCode having such excellent remote development capabilities now however, it feels like a nicer option these days but I guess only if you really care about the benefits that brings. Agreed about reimporting libraries still being a major pain point in Python, but the "advantage" for Jupiter Notebooks is also unfortunately what leads to terrible practices and bad engineering as most non-disciplined engineers end up treating it as one giant script for spaghetti code to get the job done.
Browsing code, underlying library imports and associated code, type hinting, error checking, etc., are so vastly superior in something like Pycharm that it is really hard to see why one would give it all up to work in a Notebook unless they never matured their skillsets to see the benefits afforded by a more powerful IDE? I think notebooks can have their place and are certainly great for documenting things with a mix of Markdown, LaTeX and code, as well as for tutorials that someone else can directly execute. And some of the interactive widgets can also make for nice demos when needed.
Notebooks also make for poor habits often times and as you mentioned, having data scientists and ML engineers write code as modules or commit them via pull-requests helps them grow into being better software engineers which in my experience is almost a necessity to be a good and effective data scientist and ML engineer.
And lastly, version controlling notebooks is such a nightmare. Nor is it conducive to code reviews.
As an aside, I really wish the VSCode team did more to integrate iPython REPL more seamlessly into VSCode as that is one of the big blockers for me to using VSCode for anything Python related.
That being said, if you want to better protect yourself, you should probably be wearing a better quality mask. It's the reason why I've only ever worn KN95 or N95 masks since the start of the pandemic.
Tesla has built out amazing infrastructure to capture extensive amounts of "hard" examples from their fleet, turn them around into labeled data for training very efficiently and then utilizing simulations to further broaden the distribution of such quirky long-tail events in their training-set. In the absence of AGI, this is a very effective "brute-force" approach and they have a huge upper hand over every other player in this space.
I say all of this even though I am very skeptical that anyone will achieve L5 self-driving with where the state of things are today. But Karpathy and team are very pragmatic and making lots of good decisions coupled with excellent engineering and infrastructure development.