dree-max/umi-sight

UMI Insight — AI-powered assessment and learning intelligence platform for the Uganda Management Institute Innovation Challenge. Transforms participant feedback into evidence-backed insights, improvement actions, and measurable assessment-quality outcomes. # UMI Insight — MVP Prototype **Reimagining Assessment Through Smart Feedback** — a working prototype for the Uganda Management Institute Innovation Challenge. It turns participant feedback into evidence-backed, human-validated insight for assessment improvement. Built with **Node.js (>= 22.13) and the standard library only** — zero npm dependencies, SQLite (via `node:sqlite`), built-in HTTP server, and hand-rolled SVG charts. All data is **synthetic demo data**. ## Quick start ```powershell # 1. Seed the database from the synthetic dataset (first time) npm.cmd run seed # 2. Start the server npm.cmd start # 3. Open # localhost ``` Or run the guided demo: ```powershell npm.cmd run demo ``` > On Windows, `npm.ps1` may be blocked by the execution policy — use > `npm.cmd` or invoke node directly: `node server/index.js`. ## Demo accounts (password: `demo123`) | Account | Role | Scope | | --- | --- | --- | | `qa@demo.umi.ac.ug` | Quality Assurance | Institution-wide | | `executive@demo.umi.ac.ug` | Executive | Institution-wide | | `lecturer@demo.umi.ac.ug` | Lecturer | MBA modules M001–M004 only | | `coordinator@demo.umi.ac.ug` | Programme Coordinator | MBA programme P001 only | | `admin@demo.umi.ac.ug` | System Admin | Institution-wide | | `P-0001` (or `U0001`) | Participant | MPP programme P004 | | `MROLE-001` | Participant + Coordinator | Multi-role selector demo | Participants from the dataset (360 accounts, e.g. `U0002`…`U0360`) also log in with `demo123`; they are scoped to their own programme only. ## The demo story (seeded from the synthetic dataset) 1. **Assessment clarity collapses** in Strategic Management (M001), Public Policy Analysis (M005) and Project Planning (M017): - Clarity ≥ 4 share: **83.7% (CY01) → 19.6% (CY03) → 33.9% (CY04)** - Unclear-instruction mentions in CY04: **50.8%** (30 of 59 responses) 2. Insight **INS001** (validated, high priority) captures this with a full evidence table. **INS002** captures uneven feedback …