theSage

We use jayporeci.in and have never hit capacity problems since we can choose to run it in laptops / servers / spare VMs etc.

Best decision we ever made

I've been working in ML/Data science and building websites for ~7 years now and am now slow moving into full time freelance work.

  Location: Jaipur, India
  Remote: Yes (UTC +5:30)
  Willing to relocate: No
  Technologies: Python, Django fullstack
  Résumé/CV:https://arjoonn.com/cv
  Email: arjoonn@midpathsoftware.com

I use KeepassXC + Syncthing.

On a few devices that never connect to the same wifi i use tailscale to connect Syncthing.

https://www.jayporeci.in/

I've been working on this CI system for a while.

1. Zero setup. Works on git hooks.

2. Python as the config language. Makes it very easy to do dependencies/matrix jobs/conditional jobs.

3. Offline first. It can work online with gitea as well.

4. Everything is in git. I don't need to muck around in and configure the CI system itself.

I've been building Jaypore CI for exactly this tradeoff. The config is a single python file and I run jobs using a git hook on my own laptop.

I'd love some feedback on what else I could add to this project to make life easier for people.

https://www.jayporeci.in/

Hi HN,

I built this a while back for small projects where I wanted to build a static site without introducing the whole npm/js ecosystem.

The main advantage is having re-usable parts for the HTML/CSS/JS instead of writing duplicated markup.

I usually couple this with htmx.org when I want to introduce flask/django/fastapi in the project and things work out wonderfully since I'm already in the python ecosystem.

What do you use then? Gitlab? Personal servers?

I built our local python community's website and to be honest I have no idea about what all is to be considered while designing a page.

The only thing I kept in mind was that almost all our members were using really bad 2g connections and so the site had to be super lightweight.

Anything you can help with/tell us to improve would go a long way.

https://www.pyjaipur.org/ https://github.com/PyJaipur/PyJaipur/tree/master/website

I'm not sure if the equation is right but I've done a daily case plot for Indian cases.

https://gist.github.com/theSage21/29608c666a378dcb276a9dd9e4...

For what it's worth, at my previous place we built a YOLO based model for detecting paragraphs/tables/headlines/page layouts mixed with traditional rule based OCR/layout detection.

https://www.youtube.com/watch?v=VVdHFqhQRUk

https://voody.clapresearch.com/

We use tmux for remote peer programming in a pinch.

If Alice has a session on a remote machine you have access to. You can run `tmux new-session -s bob -t Alice`

With this both share he same view but can move cursors independently.

I've been using type hints with my bottle projects for some time now. Makes it a lot easier to write JSON based apis.

https://github.com/theSage21/bottle-tools

I never thought of this. Thank you.

Hmm, perhaps I should have said "How to investigate if something went wrong?".

You're right that people leaving is normal. What I'd like to be sure of is that they are leaving because of the right reasons.

I'm their coworker and this is what I'm afraid of. I am haunted by the question "what if it has been our work environment that caused the move?"

I don't suppose toxic people think/know that they are toxic, so how do I find out if we are?

I don't think I fully understand what you've said. I'm sorry for butchering this but, do you mean to say that when specific questions are asked the underlying assumptions are so out of sync that you cannot formulate an answer they would understand?

I think an example would help?

I understand. Maybe if the conversation takes that tone we'll just have a nice tea and go back to life.

I don't understand what IC's means. Is it a standard term?

I've been using the koolertron split mechanical keyboard for a year now and it's been pretty good so far. I can sit up straighter, my thumbs and pinky don't hurt anymore.

The connector cable in the middle is becoming a little loose and it did not ship with a spare. Would have been easy enough to replace but I can't find any micro-to-mini usb cables. I'm just hoping to splice one together myself once it goes kaput.

https://www.amazon.com/Koolertron-Programmable-Mechanical-Ke...

Location: Trivandrum, India

Remote: yes

Willing to relocate: yes

Technologies: python,pytorch,svelte,ansible,jupyter

Résumé/CV: https://www.arjoonn.com/cv

Email: arjoonn.94@gmail.com

we never implemented the poincare paper since by then Facebook had published it's own code, but for the lorentz paper there were a few which were wrong. I'll update the repo to point to those in a bit

If you want to experiment with these kinds of embeddings, a stabler version using Lorentz embeddings instead of Poincare was written by us at work for internal use.

https://github.com/theSage21/lorentz-embeddings

we had a similar adventure implementing the next thing which was lorentz embeddings. Some equations were wrong in the paper which AFAIK have not been fixed in subsequent revisions.

@dchatterjee172 found a lot of the inconsistencies in the work.

https://github.com/theSage21/lorentz-embeddings

how do you keep track of these things for future reference? Browser history or something more involved than that?

David Beazley hosts a pretty good compiler class. I wrote my first one there.

https://www.dabeaz.com/compiler.html

I've been working for 2 years now. I've built APIs, managed databases, written ML models and dev productivity software so far. Looking for places with problems I can learn from.

  Location: Trivandrum, India
  Remote: Yes
  Willing to relocate: Yes
  Technologies: Python, AWS, SQL, Docker, PyTorch
  Résumé/CV: https://www.arjoonn.com/cv
  Email: arjoonn.94@gmail.com

just starting to read this. It reminds me of meiosis networks + stochastic delta rule where the weights were drawn from a random distribution (normal I think).

Location: Trivandrum, India

Remote: yes

Willing to relocate: yes

Technologies: python/webservers/pytorch/gitlab

Résumé/CV: https://www.arjoonn.com/cv

Email: arjoonn.94+hn20083793@gmail.com

I did some work on html2text when I started off and while the little parts did make sense, the whole library was confusing for me. I couldn't change anything without breaking tests.

On the other hand I've been tinkering with curio for a while now and it's a fresh breath compared to that.

My trouble is that I still don't understand what makes the html2text thing "bad". What particular thing there caused me to not like working with it? I'm trying to understand that.

I've been book hunting + figuring out if it's something that I did not know which would have made the code a lot easier to work with (stuff got a lot easier after I cleaned up my set theory understanding)

- https://github.com/dabeaz/curio - https://github.com/Alir3z4/html2text

Wrote a simple fizzbuzz server which brought down the time we spent interviewing freshers for internships/jobs. Since we're a small team, this had a big impact.