Get Involved

Whether you write code, define schemas, or just have sharp questions, there's a place for you here. We're building something applied and careful. This careful work benefits from many perspectives.

Contribution Areas

</>

Code Contributions

Parsers and adapters for communication systems. Metadata extraction and redaction helpers. Graph construction and traversal primitives. Validation tools and synthetic test corpora.

▢

Schemas and Definitions

Message and event schemas. Definitions of handoff, domain crossing, and termination. Assumptions and edge-case handling.

✎

Documentation and Examples

Clear explanations of existing primitives. Worked examples using synthetic or public data. Boundary cases that illustrate limits rather than successes.

✓

Review and Challenge

Challenging definitions that embed hidden assumptions. Identifying ambiguity in documentation. Stress-testing concepts against real-world scenarios.

How to Contribute

Five steps from idea to accepted change.

  • 1. Start with the Foundations

    Review the Technical Foundations and Areas of Focus sections to understand scope and terminology.

  • 2. Open an Issue or Discussion

    Use the project repository to describe what you propose, why it matters, and where it fits within scope.

  • 3. Submit a Focused Change

    Keep contributions narrow. Small, well-explained changes are preferred over broad refactors.

  • 4. Participate in Review

    Review is collaborative, not adversarial. We'll discuss all tradeoffs.

  • 5. Document the Outcome

    Every accepted change should leave behind clearer understanding than before.

Your Contributions Matter

All backgrounds are welcome here.

Expertise in captive insurance or graph theory helps but is entirely optional. If any of these describe you, we'd genuinely like your help:

  • Systems and data engineers who've wrestled with ingestion, parsing, or normalization and know where things break in data pipeline work and can readily fix.
  • Applied ML or graph practitioners who care about structure and reproducibility.
  • Researchers curious about governance dynamics, escalation patterns, or institutional behavior.
  • Practitioners in risk management, insurance, or compliance who can pressure-test our assumptions against what actually happens.
  • Anyone who wants to read the whitepapers and other materials and thinks "this part is confusing", that's a contribution too.

How We Work Together

Transparency and precision in service of regulated environments.

BBCO serves captive insurance programs where governance quality directly affects capital efficiency. That means the work eventually reaches actuaries, risk managers, and boards in regulated settings, so we aim for transparency and precision.

What good contributions look like:

  • Think about privacy and auditability early. The BBCO framework is built to handle real organizational communications. We value contributions that demonstrate how metadata can be handled with precision while minimizing exposure of underlying content. Since BBCO is a framework for processing and interpretation rather than data collection, linkage logic should remain transparent and reproducible within the contributor’s own environment.
  • Favor determinism over sophistication. Methods that produce consistent, repeatable outputs are especially valuable. When LLMs or statistical models are part of the pipeline, we encourage low-temperature settings, aggregated evaluation, explicit confidence bounds, and defined fallback paths so that variability is understood rather than assumed.

Good first contributions:

  • Browse the foundational whitepaper to see the full pipeline, from extraction through variance acceleration, and think about where your skills fit
  • Spot confusing terminology in an existing definition or schema and suggest clearer language
  • Add a worked example using synthetic or public data, the Enron corpus is a good starting point
  • Find an edge case where a method might produce misleading results and write it up, these are genuinely valuable
  • Improve extraction patterns for a specific email format (MBOX, EML, PST) based on your real-world parsing experience
Request GitHub Access