alibaba/Alink

Health v2 · maintenance only

Adoption not part of the score

3611 stars · 782 forks observed · 2026-08-28

What it is AI-extracted, prompt v1, taxonomy v1, 2026-08-29, confidence not recorded

Alink is a machine learning algorithm platform built on Apache Flink, developed by Alibaba's PAI team. It provides a large library of batch and streaming ML algorithms (classification, clustering, regression, recommendation, graph embedding) usable from Java and Python (PyAlink).

Use cases

• run machine learning pipelines on flink
• train classification and regression models on streaming data
• build recommender systems at scale
• feature engineering on big data
• distributed clustering and graph algorithms
• use xgboost with flink
• word2vec and graph embedding on flink

When to choose

• you already use Apache Flink and want ML on the same platform
• you need both batch and streaming machine learning
• you want a broad algorithm library with Java or Python APIs

When to avoid

• you don't run Flink and don't want a JVM/Flink dependency
• you need a lightweight single-node scikit-learn-style workflow
• you need cutting-edge deep learning training rather than classic ML

Facets

framework · maturity maintenance

machine-learning etl streaming data-science nlp machine-learning data-science big-data jvm python cross-platform flink apache-flink distributed-ml feature-engineering recommender-systems xgboost pyalink alibaba data-engineering

Member repositories

For agents

markdown · JSON · MCP: product_card(name="alibaba/Alink")