Better Dev Link

Have you ever feel overwhelm with all the news about LLM? what is these 70B 4B or 3.5 4.0? Or when you go and check out the code of some project, it’s super short and simple, where is the logic? In this link, we go back to that root to understand how to build an LLM by using Pytorch.

Have you ever feel like iCloud drive, Google Drive or Dropbox app is magic? What power them? It’s pretty much FUSE behind the scene. Can we turn random data available through RPC into a file system?

Cloud storage is ubiquitous: Google Drive, Dropbox, and OneDrive are household names. However, these services do not provide end-to-end encryption (E2EE), meaning that the provider has access to the data stored on their servers. The promise of end-to-end encrypted cloud storage is that users can have the best of both worlds, keeping control of their data using cryptographic techniques, while still benefiting from low-cost storage solutions.

QRCode is everywhere nowsaday. However, they doesn’t need to be just square black and white box. How can we generate QRcode that looks artsy while still readable to the decoder?

With a small team and in a few monthes, imblue trained a 70B parameter model from scratch on their own infrastructure that outperformed zero-shot GPT-4o on reasoning-related tasks. They’re sharing an end-to-end guide for setting up the required infrastructure: from bringing up the initial cluster and installing the OS, to automatically recovering from errors encountered during training.

what if I told you that simply reordering your columns could reduce the size of your tables and indexes by 20%? This isn’t some obscure database trick — it’s a direct result of how Postgres aligns data on disk.

This article is a look at how ngrok, a platform to enable you expose local service on your laptop to internet, built their data platform. Given their team size of 1 engineer, this maybe more useful and relevant to us more than scale of big corp(where there are hundred engineers building a data platform).

Threads and the general concept of concurrency are among the most widely misunderstood and confusing topics for newer developers. This series is meant to help clarify some of these concepts and provide a better understanding of how they work.

a very high level discussion of how Netflix improve and evolve their websocket infrastructure to perform many kind of tasks. There isn’t much technical discussion on this post, but more about concept and system design.