---
title: "AI GPU Clusters, From Your Laptop, With Livebook"
description: "Livebook, FLAME, and the Nx stack combine Elixir notebooks, elastic execution, and GPU-backed machine learning into a workflow that can move from a laptop to remote infrastructure. Livebook can start runtimes on Fly Machines, connect to existing Erlang/Elixir applications, and synchronize notebook dependencies and modules across nodes while preserving access to private resources within a Fly organization. FLAME wraps arbitrary code in Flame.call, manages executor pools, controls node limits and concurrency, streams results, and shuts down idle nodes or the cluster after disconnection. In the demonstrated AI workflow, video stills are processed on GPU Fly Machines with Llama, then Mistral summarizes the returned descriptions. The same setup also drives a 64-GPU hyperparameter run for BERT, with each L40s node preparing dependencies, loading data, fine-tuning, and streaming chart results to Livebook."
---

# AI GPU Clusters, From Your Laptop, With Livebook

[Fly.io](https://yomu.fyi/company/fly-io) · Sep 24, 2024

**Type:** Explainer

## Summary

Livebook, FLAME, and the Nx stack combine Elixir notebooks, elastic execution, and GPU-backed machine learning into a workflow that can move from a laptop to remote infrastructure. Livebook can start runtimes on Fly Machines, connect to existing Erlang/Elixir applications, and synchronize notebook dependencies and modules across nodes while preserving access to private resources within a Fly organization. FLAME wraps arbitrary code in Flame.call, manages executor pools, controls node limits and concurrency, streams results, and shuts down idle nodes or the cluster after disconnection. In the demonstrated AI workflow, video stills are processed on GPU Fly Machines with Llama, then Mistral summarizes the returned descriptions. The same setup also drives a 64-GPU hyperparameter run for BERT, with each L40s node preparing dependencies, loading data, fine-tuning, and streaming chart results to Livebook.

## Context

The keynote recap presents Livebook, FLAME, and the Nx stack as Elixir components for local-to-remote computation, elastic infrastructure, and AI or machine-learning workflows. Examples include processing video archives and tuning BERT across GPU nodes from a laptop.

## Approach / What changed

Livebook starts runtimes on Fly Machines or connects to running Erlang/Elixir applications. FLAME manages pools of executors and dispatches notebook code across ephemeral nodes, while Nx, Axon, and Bumblebee provide tensor computation and model support. The demonstrations use Llama and Mistral for video descriptions and summaries, then scale BERT hyperparameter tuning across 64 L40s GPU machines.

## Takeaways

- Livebook can connect a laptop notebook to Fly Machines and existing Erlang/Elixir applications, including services on private networks within a Fly organization.
- FLAME synchronizes notebook dependencies and code across nodes, lets users set minimum and maximum node counts and concurrency, and automatically replaces completed work until processing finishes.
- A demonstrated BERT hyperparameter run used 64 Fly Machines with L40s GPUs, streaming each node’s fine-tuning results to charts in Livebook.

**Tags:** [Elixir](https://yomu.fyi/topic/elixir), [Fly Machines](https://yomu.fyi/topic/fly-machines), [GPU](https://yomu.fyi/topic/gpu), [Livebook](https://yomu.fyi/topic/livebook), [Machine Learning](https://yomu.fyi/topic/machine-learning)

- Source: [Fly.io](https://fly.io/blog/ai-gpu-clusters-from-your-laptop-livebook)
- Source URL: https://fly.io/blog/ai-gpu-clusters-from-your-laptop-livebook
- Ingested by Yomu: 2026-09-03T01:26:22.275Z

[Read original post](https://fly.io/blog/ai-gpu-clusters-from-your-laptop-livebook)