Configuration playground for domain experts

Context

Loan assessment logic changes often start with analysts who understand credit policy but cannot safely experiment in production. Ad-hoc scripts and shared spreadsheets do not scale review, and every engineer becomes a bottleneck for “what if we changed this threshold?”

Constraints

• No customer or production data in the playground; synthetic and bounded inputs only.
• Changes must be explainable to compliance-minded reviewers, not only to the author.
• The tool should complement — not replace — the authoritative servicing systems.

Approach

1. Separate policy exploration from execution — configurable scenarios with explicit inputs and expected outputs analysts can diff.
2. Make boundaries obvious — which parameters are editable, which are frozen, and which require engineering release.
3. Prefer inspectable defaults — saved scenarios become the regression set when policy shifts return next quarter.

Trade-offs

A playground is not a full staging clone; some integration behaviors stay mocked. That is intentional: the goal is faster policy reasoning, not pretending the entire core bank runs on a laptop.

Outcomes

• Analysts iterate on assessment questions without queueing engineering for every variant.
• Engineering receives clearer, test-shaped requests when a scenario is ready to promote.
• The same “bounded environment” instinct later informed gateway and promotion tooling work on the integration edge.