Self hosting and Log sharing

I got my hands on some really cool self-hosting tech today, Coolify. If you didn't know, there are great services like Vercel that allow you to easily deploy applications, from your github repository. Well, Coolify is that, but your own the server.

As i'm trying to share more of my work and have the most control, i decided to take the Coolify self hosting path, although they do have a cloud.

Now that I have my platform setup, i deployed Langfuse. With this deployment I can share different traces and generations that I create with the world. This means more transparency to how i'm rating these LLMs, which is great for the broader AI/ML community for picking models they think can perform on particular task.

Check out a simple trace here.

Also, I can host my silly demos that i build to display the data, by simlpy giving my Coolify deployment the link to my Private Github repo.

Check out a silly demo here.

Is self hosting for everyone? Probably not, but I think it makes you a more well rounded practitioner knowing how to at least setup your own Heroku like platform on top of your own compute.

3 comments