FLI Scorecard Operations System

Client: Frontera Leadership Institute (FLI) - Ruben Cantu
Built by: Alan Salinas - Caelus Consulting
Deadline: ~May 8, 2026
Status: Week 5 ready - testing, private deployment, and delivery

---

What This Is

A competency-based evaluation system for FLI's 11-student Spring 2026 cohort. Transforms Google Form evaluation exports into:

• Interactive React dashboard (3 screens) for staff
• Per-student PDF credentials for employers and BCIC
• Cohort impact report for institutional stakeholders
• Standalone AI Self-Check tool for student self-calibration

See .claude/CLAUDE.md for full architecture, schema, and tech decisions.
See docs/week-5-testing-deployment.md for final QA, private deployment, and PDF delivery steps.

---

Repo Structure

fli-scorecard/
|-- etl/           Python ETL - run locally to process .xlsx to JSON
|-- app/           React dashboard - deploy to Vercel
|-- ai-self-check/ Standalone AI Self-Check prompt builder (no API)
`-- docs/          Build specs, prompts, and delivery runbooks

---

Quick Start

Step 1 - Run the ETL

After evaluations are exported from Google Sheets:

cd etl
pip install -r requirements.txt
python etl.py path/to/evaluations.xlsx
# Outputs: cohort_data.json
cp cohort_data.json ../app/src/data/cohort_data.json

For development, use etl/sample_data/sample_cohort.xlsx. Real student workbooks and real generated JSON must stay out of git.

Step 2 - Run the React App

cd app
npm install
npm run dev
# http://localhost:5173

Step 3 - Deploy to Vercel

Use Vercel project root directory app.

• Build command: npm run build
• Output directory: dist
• Framework preset: Vite

For real student data, prefer a local Vercel deploy with the validated, uncommitted app/src/data/cohort_data.json. GitHub auto-deploy should only use sample/anonymized data unless Alan approves a different real-data handling process.

---

Data Flow

Google Forms -> Google Sheets -> .xlsx export
     |
etl/etl.py -> cohort_data.json -> app/src/data/ -> React dashboard

Security: Real student evaluation data never enters the repo. Use etl/sample_data/sample_cohort.xlsx for development. cohort_data.json is gitignored.

---

PDF Export

• Screen 2 (Student Deep Dive) -> "Export PDF" -> student credential one-pager
• Screen 3 (Cohort Analytics) -> "Export Cohort Report" -> BCIC-ready document

AI Self-Check

Open ai-self-check/index.html directly in a browser. It builds a copyable Claude prompt locally; it does not call Anthropic APIs, require a key, or store student data.

---

Build Tracker

See .claude/CLAUDE.md -> Build State Tracker section for week-by-week task checklist, and docs/week-5-testing-deployment.md for the final delivery checklist.