Evaluate business rules on millions of records - a native PySpark integration for your clusters, and the Rust engine for throughput-critical pipelines.
Embedded in the job, the engine sweeps your whole book without a single per-row API call - same rules, same answers, hours faster.
Rules behind an API, called per row28 hours
Engine embedded in the job4 minutes
10M rows, same rules, same answers - the network hop is the whole difference.
1M+evaluations per second with the Rust engine
< 1 msaverage evaluation latency
10xfaster than interpreted engines
Choose your runtime.
Two ways to process rules at scale, depending on where your data lives and how fast it has to move.
PySpark
For Spark clusters
Native PySpark integration for evaluating rules across distributed datasets. Works with Databricks, EMR, Dataproc, or self-managed Spark clusters.
Distributed processing across workers
DataFrame API integration
Works with existing Spark jobs
Python-native development
pip install zen-engine
Rust SDK
For custom pipelines
Maximum throughput with the native Rust engine. Build custom data pipelines or embed it in existing Rust applications for throughput-critical workloads.
Millions of evaluations per second
Zero-copy memory management
Multi-threaded processing
Minimal resource footprint
cargo add zen-engine
Built for the jobs that touch everything.
Common patterns for rule evaluation at scale - anywhere the row count has too many zeros for an API.
Batch processing
Run millions of records through your rules - nightly jobs, migrations, bulk calculations.
Analytics pipelines
Classify, score, or transform data inside your ETL before it lands in the warehouse.
Streaming data
Evaluate rules on Spark Streaming as events arrive, not after the fact.
Backfill operations
When rules change, re-evaluate historical records with the updated logic in one pass.
Ready to process at scale?
Start with the open-source SDKs and documentation, or talk to us about your pipeline.
