My question is that how will you deal with future government regulations. In developing world, the government does not like crypto and may impose laws like upfront taxation to discourage people from investing.
While we want to have constructive conversations with some of these governments (I actually believe it's possible), ultimately we can't stop hostility / regulations against crypto. What we can do is offer our service and be transparent about what it is — and then users in those countries can make decisions for themselves about whether they can/should engage.
I'm hopeful, however, seeing countries like Turkey and a couple in LatAm that have significant crypto adoption without overt government hostility — when adoption reaches a critical mass, I expect to see more governments relax restrictions.
ML is mostly Python territory these days, and for good reason—it’s simple, flexible, and gets the job done. But as someone who spends most of their time in C++, I thought: What if I tried building something ML-related in C++? Sure, I had an idea of the challenges — verbose syntax, manual memory management, and, let’s face it, a lot less “fun” compared to Python—but I wanted to see what it’s really like.
So, inspired by micrograd, I decided to build my own autograd library in C++ from scratch. This wasn’t about translating Karpathy’s code (I’ve barely read it, honestly). My goal was to understand the inner workings of Autograd and Computational graphs, like the ones used in PyTorch, and recreate those ideas in C++.
The result? A small library that can train simple neural networks for tasks like linear and logistic regression.
It’s been a wild ride—balancing the elegance of ML concepts with the quirks (and occasional frustrations) of C++.
If you’ve ever wondered what it’s like to tackle ML with C++, or just enjoy geeking out over low-level implementations, I’d love to hear your thoughts. Code and project details are here: https://github.com/apexkid/micrograd-cpp.
I took on a new year's resolution to resume writing blogs. The last I did was when I was doing my undergrad. At that time, blogging was not that popular but I still loved it. But I got busy in work and lost that passion.
So here is to new beginning. Looking for support from the community.
I am working on Zipshot. http://zipshot.co -- The goal is to make the best free screenshot & OCR app for mac.
This is my side project because I needed an app like this for my daily use and I hate the fact that every other good app is paid even though maintaining such an app really doesn't cost anything.
Zipshot solves several problems:
- Easy to use shortcuts like cmd+shift+1
- No desktop clutter as images are uploaded to Zipshot cloud.
- Share via links.
- Powerful editor and figma style comments for annotations.
- OCR that works without the internet on 11 languages.
- Slack & Gmail integration.
- Context aware naming.
- Serious privacy measures.
- One click data download.
Who is it for:
- Users who take 5+ screenshots a week with mostly the purpose of sharing it with their team or friends.
I am improving it every week. Here are the things I am working on:
- Make the on-boarding and permission seeking workflow simpler.
- Speed up the app for apple silicon devices.
- Support OCR for rotated text and include more languages.
> periodically just take a snapshot of everything in RAM.
Sound similar to `stop the world Garbage collection` in Java. Does your entire processing comes to halt when you do this? How frequently do you need to take snapshots? Or do you have a way to do this without halting everything
Good catch! Snapshotting was certainly a bottleneck that I chose not to write about.
But we aren't really taking the snapshot of RAM, instead we're running some code asking each object to snapshot itself into a stream. If you do this naively, it will block writes on the server until the snapshot is done (reads will continue to work).
But Raft has a protocol for asynchronous snapshots. So in the first step we take an immutable fast snapshot of the state we care about which happens quickly, then writes can keep going while in the background we serialize the state to disk.
I like your vision but I do not understand the fundamental proposition here. The creators in the platform are supposed to perform specific tasks to raise money -- That's called "having a job". Essentially the platform is a marketplace to hire part time workers for specific one time tasks. If so, why position it as a campaign to raising funds as if it is some sort of charity.
Is the underlying assumption that creators can charge more on this platform as compared to the usual pay they will get for the same task and in doing so they can raise money faster? If so, I would be highly skeptical of this assumption and validate it rigorously.
Hmmm... I don't see it as a charity. Campaign creators create their campaigns and offer their services. People in their community can buy those services. Let's say a person who has a full-time job, a person you know, wants to raise $1000 to help pay her rent and offers babysitting services in return. You need this service and instead of searching for someone you don't know, you hire that person. At the same time, you get a job done and you know you did a good things and helped her pay her rent.
Unless you were recently hired, you should look out for another job and quit. The argument that you are already compensated as per market standards only holds true if you were hired less than an year ago. Also, the company should have given you raise even if it was smaller value in percentage terms. Not giving any raise is a clear indication.
My question is that how will you deal with future government regulations. In developing world, the government does not like crypto and may impose laws like upfront taxation to discourage people from investing.