Update README with bucket pipeline docs and benchmarks

Reorganize to lead with the new bucket-based pipeline (Jobs + Buckets +
Spaces) as the recommended approach. Add performance benchmarks, flow
diagram, and prep step examples. Move static export scripts to legacy section.

Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>

README.md

@@ -1,30 +1,105 @@
 ---
 viewer: false
-tags: [uv-script, hf-jobs]
+tags: [uv-script, hf-jobs, hf-buckets]
 ---
 
 # Atlas Export
 
-[Embedding Atlas](https://github.com/apple/embedding-atlas) is an open-source library from Apple for creating interactive, browser-based visualizations of embedding spaces. It renders millions of data points with WebGPU acceleration, supports real-time search and filtering, and automatically generates cluster labels — all running as a static site with no backend.
+[Embedding Atlas](https://github.com/apple/embedding-atlas) is an open-source library from Apple for creating interactive, browser-based visualizations of embedding spaces. It renders millions of data points with WebGPU acceleration, supports real-time search and filtering, and automatically generates cluster labels.
 
-These scripts wrap Embedding Atlas to make it easy to go from a HuggingFace dataset to a deployed visualization in a single command. See [arxiv-cs-atlas](https://huggingface.co/spaces/davanstrien/arxiv-cs-atlas) for a live example (215K arXiv CS papers).
+These scripts wrap Embedding Atlas to make it easy to go from a HuggingFace dataset to a deployed visualization. See [open-library-atlas](https://huggingface.co/spaces/davanstrien/atlas-bucket-test) for a live example (2M books).
 
 ![Example Visualization](https://huggingface.co/datasets/huggingface/documentation-images/resolve/58e8cf44ec5d74712775b7049b1203f9b0b15ebb/hub/atlas-dataset-library-screenshot.png)
 
 ## Scripts
 
-This repo contains two scripts:
-
 | Script | Description | Best for |
 |--------|-------------|----------|
-| `atlas-export.py` | All-in-one: embeds data into the Space | Small/medium datasets (<10GB) |
-| `atlas-export-remote.py` | Splits data into a separate HF dataset repo | Large datasets (10GB+), images |
+| **`atlas-e2e.py`** | End-to-end: dataset to deployed Space (experimental) | One-command pipeline |
+| **`atlas-build-gpu.py`** | GPU atlas build with cuml.accel UMAP | Running as an HF Job |
+| **`atlas-deploy.py`** | Deploy a Space from bucket data | Deploying existing builds |
+| `atlas-export.py` | All-in-one static export | Small datasets (<10GB) |
+| `atlas-export-remote.py` | Static export with remote data | Medium datasets |
+
+## Bucket Pipeline (Recommended)
+
+The new pipeline uses [Storage Buckets](https://huggingface.co/docs/hub/storage-buckets) + [Jobs](https://huggingface.co/docs/hub/jobs) + [Spaces](https://huggingface.co/docs/hub/spaces) for large-scale visualizations:
+
+```
+atlas-e2e.py (local orchestrator)
+  │
+  ├─ 1. Creates Storage Bucket
+  │
+  ├─ 2. Submits GPU Job (atlas-build-gpu.py)
+  │     • Embeds text/images on GPU
+  │     • UMAP via cuml.accel (~50x faster)
+  │     • Writes parquet to bucket
+  │
+  └─ 3. Deploys Docker Space (atlas-deploy.py)
+        • Mounts bucket (read-only)
+        • Serves embedding-atlas in server mode
+        • Server-side DuckDB (not browser WASM)
+```
+
+### Quick Start
+
+```bash
+# One command: dataset → deployed Space
+uv run atlas-e2e.py stanfordnlp/imdb \
+    --text text --split train \
+    --name imdb-atlas --sample 50000
+```
+
+### Step by Step
+
+```bash
+# 1. Build atlas on GPU (runs as HF Job)
+hf jobs uv run --flavor a100-large \
+    -v hf://buckets/user/atlas-data:/data \
+    -s HF_TOKEN --timeout 2h \
+    atlas-build-gpu.py my-org/my-dataset \
+    --text text --name my-atlas --sample 1000000
+
+# 2. Deploy Space from bucket
+uv run atlas-deploy.py \
+    --name my-atlas \
+    --bucket user/atlas-data \
+    --space-id user/my-atlas-viz
+```
+
+### With a Prep Step (for filtering/enrichment)
+
+```bash
+# Prep: filter and add categories with DuckDB
+hf jobs uv run --flavor cpu-upgrade \
+    -v hf://buckets/user/atlas-data:/output \
+    -s HF_TOKEN \
+    open-library-prep.py --output /output/books/books.parquet
+
+# Build: embed the prepped data
+hf jobs uv run --flavor a100-large \
+    -v hf://buckets/user/atlas-data:/data \
+    -s HF_TOKEN --timeout 2h \
+    atlas-build-gpu.py /data/books/books.parquet \
+    --text title --name books-atlas --sample 2000000
+
+# Deploy
+uv run atlas-deploy.py --name books-atlas --bucket user/atlas-data
+```
+
+### GPU UMAP Performance
+
+The build script uses [cuml.accel](https://docs.rapids.ai/api/cuml/stable/cuml-accel/) for zero-code-change GPU acceleration of UMAP. No changes to embedding-atlas needed — just an environment variable.
 
-### Why remote?
+| Dataset | Rows | Time (A100) | Cost |
+|---------|------|-------------|------|
+| IMDB | 5K | ~1 min | $0.04 |
+| TinyStories | 1M | ~20 min | $0.83 |
+| Open Library | 2M | ~40 min | $1.67 |
 
-`atlas-export-remote.py` stores the parquet data in a separate HF dataset repository and the viewer in a lightweight Space. The viewer loads data on-demand via HTTP range requests — no storage limits on the Space side. This uses embedding-atlas >= 0.18.0's native `--export-metadata` for clean remote data loading.
+For comparison: CPU UMAP on 250K rows took >2 hours.
 
-## Quick Start
+## Legacy Scripts
 
 ### Basic (data embedded in Space)
 
@@ -32,7 +107,7 @@ This repo contains two scripts:
 uv run atlas-export.py stanfordnlp/imdb --space-name my-imdb-viz
 ```
 
-### Remote data (recommended for larger datasets)
+### Remote data
 
 ```bash
 uv run atlas-export-remote.py stanfordnlp/imdb \
@@ -215,13 +290,14 @@ Run either script without arguments to see all options.
 
 ## Live Examples
 
+- [open-library-atlas](https://huggingface.co/spaces/davanstrien/atlas-bucket-test) — 2M books with category coloring (bucket pipeline)
+- [nemotron-v3-atlas](https://huggingface.co/spaces/davanstrien/nemotron-v3-atlas) — 250K NVIDIA training data samples
 - [arxiv-cs-atlas](https://huggingface.co/spaces/librarian-bots/arxiv-cs-atlas) — 215K arXiv CS papers (lance format)
-- [finephrase-faq-atlas](https://huggingface.co/spaces/davanstrien/finephrase-faq-atlas) — 50K FAQ entries (glob pattern)
 
 ## Credits
 
-Built on [Embedding Atlas](https://github.com/apple/embedding-atlas) by Apple (>= 0.18.0 for remote data support). See the [documentation](https://apple.github.io/embedding-atlas/) for more details.
+Built on [Embedding Atlas](https://github.com/apple/embedding-atlas) by Apple (>= 0.19.1). GPU UMAP via [cuML](https://docs.rapids.ai/api/cuml/stable/cuml-accel/).
 
 ---
 
-Part of the [UV Scripts](https://huggingface.co/uv-scripts) collection 🚀
+Part of the [UV Scripts](https://huggingface.co/uv-scripts) collection