What I built
I built an independent Python and SQLite budget simulator using three invented medical-school funds. It projects remaining balances, flags faculty effort above 100% across funds, and checks supplied allowability and award-period flags before applying a predefined reallocation scenario.
The example connects budget projections to the underlying records: an investigator’s combined effort is 115%, one fund projects a $41,000 deficit, and two charges carry adverse cost or award-period flags. The reallocation shows how a chosen set of adjustments changes those findings.
Explore the grant & budget simulator
The browser walkthrough models three synthetic medical school funds (NIH-R01-URO, DOD-BIONICS-26, and MED-URO-OPS) and surfaces four budget and rule findings with a combined illustrative amount of $72,800. Visitors can inspect each fund’s projections, then click “Apply example reallocation” to view predefined charge transfers, a burn-rate adjustment, and an investigator effort change that clear this scenario’s checks.
All funds, investigators, award numbers, and ledger charges are invented fixtures.
Architecture and decisions
- Evaluate budget burn and supplied cost flags together. The evaluator combines monthly burn projections with explicit allowability and award-period flags; it does not determine legal allowability from the charge description.
- Aggregate FTE effort across every funding source. Rather than inspecting grants in isolation, the personnel check sums committed effort percentages across federal awards and clinical operating accounts to catch >100% FTE over-commitments early.
- Make the scenario adjustments explicit. A predefined transfer moves flagged charges and $8,500 of monthly burn to the operating reserve and sets one investigator’s grant effort to 25%; it demonstrates a rule-passing reallocation under the simulator’s checks.
My contribution
I defined the multi-fund data model, supplied-flag and effort checks, and scripted reallocation—drawing on experience in federal A-133 compliance auditing at the Texas State Auditor’s Office and research operations at Cochrane Urology—using the AI-assisted approach described on the About page.
Scope and verification
Local validation passed 3 unit tests covering the supplied scenario’s four finding types, its resolution after predefined adjustments, and SQLite persistence. All records are invented, and the example does not establish comprehensive grant compliance or sponsor approval. The pinned source revision is available locally and has not been published to GitHub.
Source availability
The reviewed implementation is not publicly available yet. Source publication will follow a separate review; use the synthetic browser walkthrough above in the meantime.
Source snapshot 23f4ec3 · Reviewed 2026-10-25. This case study describes that revision; development may continue separately.