Unifies multimodal data, intelligence, and continuous learning with Python and SQL interfaces, seamlessly scaling from local environments to Ray clusters.
import vane vane.configure(runner="ray") con = vane.connect() assets = con.sql(""" SELECT asset_id, uri, media_type FROM read_parquet('s3://raw-assets/*.parquet') WHERE media_type IN ('image', 'video', 'audio') """) features = assets.map_batches( DecodeAndInfer, # user UDF; 1 model load/actor schema=feature_schema, # explicit user schema gpus=1, actor_number=4, ) features.write_parquet("s3://model-ready/features/")
From multimodal model training to enterprise data pipelines, real-world AI runs on diverse data. Pick the pipeline that matches your workload.
Turn images, video, audio, documents, tables, and sensor logs into filtered, labeled, deduplicated training dataset releases — with lineage and reproducible runs.
Explore →Turn PDFs, images, video, logs, forms, spreadsheets, and documents into auditable facts and agent-ready context — in SQL.
Explore →Benchmarking multimodal AI pipelines across audio, video, document, and image workloads.
36 CPU cores · 64 GB RAM2080 Ti (modified VRAM) · 22 GB GPU memory
This single-node evaluation adapts the workload design from the Ray Data multimodal AI benchmark. Daft's OOMs on the image and audio workloads may reflect the test machine's limited memory, while the older 2080 Ti GPU also limits throughput. Given current hardware and compute-budget constraints, we have not reproduced the original benchmark's full cluster-scale setup; these results apply only to the recorded single-node environment.
Unifies data, models, and agents — enabling continuous learning and scalable execution from local devices to Ray clusters.
Unified multimodal data from any source.
Train, evaluate, and align models across modalities.
Act in the real world, solve tasks, create value.
New data, outcomes, and feedback continuously improve the system.
Sensors, metadata, lineage, and model artifacts under one execution semantics.
Continuous flow for large objects with adaptive batching and pressure control.
CPU, GPU, IO, and model inference overlap through asynchronous scheduling.
The same pipeline runs across local devices and Ray clusters.
pip install vane-ai
Start from the docs examples and adapt the pipeline to your data.
Use the docs examples and llms.txt files to wire Vane into your stack.