The High-Performance, In-Memory Java Calculation Graph
Turn any Excel workbook into a scalable, in-memory calculation object.
HyperCell is a specialized Java library designed to solve a specific, painful enterprise problem: Running complex Excel business logic in high-scale backend environments.
It is not just a file parser (like Apache POI). It is a Headless Spreadsheet Engine that hydrates Excel models into optimized Directed Acyclic Graphs (DAGs) for sub-millisecond calculation.
Business analysts model the world in Excel. Engineers rewrite that logic in Java/Python. This translation layer causes bugs, delays, and drift.
HyperCell changes the workflow:
- Model: Analysts build sophisticated logic in Excel (Financial models, Risk raters, Pricing engines).
- Hydrate: HyperCell loads the workbook into memory, compiling formulas into an executable dependency graph.
- Execute: The backend injects inputs, triggers calculation, and reads outputs.
- Scale: Serialize the compiled model and distribute it across thousands of server nodes.
- In-Memory Performance: Optimized for calculation speed, not file I/O.
- Dependency Graph Resolution: Intelligent recalculation (only updates what changed).
- Full Formula Support: Implements standard Excel library (
SUM,VLOOKUP,INDEX/MATCH,IF,NPV,IRR). - Extensible Registry: Plugin architecture to register custom Java functions callable from Excel (e.g.,
=MY_DB_LOOKUP(A1)). - Serialization: Native support for freezing compiled models to binary for distributed caching.
- Thread-Safe Execution: Designed for high-concurrency server environments.
(Coming soon - Artifacts will be published to Maven Central)
implementation 'io.hypercell:hypercell-core:0.1.0'- Vision & Philosophy: Why HyperCell exists and the problem it solves.
- Use Cases: Detailed examples of HyperCell in production.
- Architecture: How the calculation graph works under the hood.
HyperCell is the open-source calculation kernel extracted from Scoop Analytics. It powers the "Digital Data Analyst" platform, handling millions of formula evaluations for autonomous data investigation.