I've been working on a project called exosphere.host. It's a platform focused on running batch AI agents at scale. It’s built for handling large data, supporting open source models, and plugging into common tools and workflows easily. We’re also working on Orbit, our orchestration engine, which will be open sourced soon.
I'm hoping to get community thoughts on a few things: - What are some real-world use cases where you’ve used batch inference or background AI jobs? - Are there any tools or services you often connect with in those setups (e.g. storage, queues, databases)? - What pain points have you faced when scaling AI workflows across large datasets? - How do you currently schedule, retry, or monitor long-running AI tasks? - Do you prefer cloud-hosted solutions or self-hosted orchestration for these workloads? Why? - If you could snap your fingers and fix one thing in your current AI pipeline, what would it be?
Would love to feature community use cases and feedback as we improve the platform. Happy to chat if anyone wants to try it out!