Snorkel AI
Snorkel AI is a data-development / AI data platform built on the open-source Snorkel weak-supervision library. The enterprise Snorkel Flow platform supports programmatic labeling, data curation, model training, and evaluation, and is driven programmatically through a Python SDK that targets a per-instance REST API rather than a single public hosted endpoint. Snorkel also delivers Expert Data-as-a-Service for frontier-model training and evaluation data.
APIs
Snorkel Flow SDK / Platform
The Snorkel Flow Python SDK (snorkelflow) drives the platform programmatically, connecting to a customer's Snorkel Flow instance via SnorkelFlowContext and an API key. The snork...
Programmatic Labeling
SDK surface (snorkelflow.studio, snorkelflow.lfs, snorkelflow.operators, snorkelflow.templates) for authoring code-based and template-based labeling functions, transformation an...
Evaluation
Platform and SDK capabilities for assessing model performance across data splits, running error and slice analysis, and building expert-curated benchmarks. Largely delivered in-...
Open-Source Snorkel Library
The free, open-source snorkel Python library (Apache-2.0) for programmatically building and managing training data with weak supervision - labeling functions, the LabelModel, tr...