Self-Hosted AI in Production: An Afternoon with PineGrove AI
In recent years, open-source LLMs have gotten better and better, to the point where they can solve very real problems - all inside software and hardware you can be in control of.
But while the open-source LLM world is leaping forward at break-neck pace, keeping up with it is a full-time job of its' own - and that's before we even start to consider all the surrounding elements which actually integrate those models with your business.
At PineGrove, we've been following the chaos, moving from a small model on a Mac at home to a production-ready system running real-world models on beefy GPUs serving multiple customers. And now we're ready to share our hard-earned wisdom.
Join us for an afternoon where we'll share our stories from the trenches, and the landmines to avoid along the way.
As part of this story, we will take you through a zero to hero live-building of a product powered by open-source AI. It may be a product no one asked for - but hey, at least it's yours.
Through the afternoon, we'll ask and answer questions such as:
How do I download and run an open-source model?
How do I know which model to pick (and what hardware does it take?)
What is the actual meaning and use-case of:
A harness?
An agent?
Tools?
Skills?
Data context/ingestion/RAG?
Orchestration?
...
And finally... What useful things can I actually do with all that?
Come join us in building something useless and learn something useful!