Grant Tracker — from typewriter to database

2 CFR

200 Compliance

SF-425

Federal Reporting

RBAC

Multi-Role Access

The Problem

Educational institutions managing multi-year federal grants struggle with budget compliance, document organisation, and maintaining complete audit trails across departments. Existing solutions were either too rigid (enterprise ERP modules) or too lightweight (spreadsheets).

GrantTracker 2.0 treats compliance knowledge as governed infrastructure — not a checklist — building sponsor-aware logic directly into the workflow so teams can focus on outcomes rather than paperwork.

Key Features

✓Multi-Year Grant Tracking

✓AI-Driven Insights & Chat

✓Sponsor-Aware Compliance (2 CFR 200)

✓SF-425 Federal Reporting

✓Role-Based Access Control

✓Real-time Budget Monitoring

Architecture: Type-Safe API

The entire data layer is built with tRPC — giving end-to-end TypeScript type safety from database schema to React component without a manual API layer. Prisma handles database access with generated types that match the Postgres schema exactly.

// tRPC Router — end-to-end type safety
export const grantRouter = createTRPCRouter({
  getById: protectedProcedure
    .input(z.object({ id: z.string() }))
    .query(async ({ ctx, input }) => {
      const grant = await ctx.db.grant.findUnique({
        where: { id: input.id },
        include: { grantYears: true }
      });
      return grant;
    }),
});
          

Role of AI

Multi-agent AI powers the proactive spending insights and policy query system — users can ask natural language questions like "Are we at risk of violating the salary cap under 2 CFR 200.431?" and get grounded, context-aware answers with references to the relevant grant year data.

Access: the live demo is a frontend-only showcase on mock data — the cockpit, not the engine. The private full-stack build carries the multi-year grant, budget, audit, and multi-agent AI foundation (tRPC API, Prisma data layer, Clerk auth); the demo’s reallocation-approval and SF-425 workflows run end to end on mock data and are the design spec for the engine’s workflow and federal-reporting modules, in development. Available to discuss in a working session.