Live demo: https://construction-dashboard-dusky.vercel.app
A portfolio project demonstrating how AI and data analysis skills can turn raw construction project data into actionable operational intelligence — the kind of capability that extends Procore's value without replacing it.
Construction companies using Procore accumulate rich data: project timelines, budgets, change orders, subcontractor performance, safety records, task completion rates. Most of that data surfaces in dashboards that show what happened — not what is about to go wrong.
Project managers end up reacting to problems rather than preventing them. A project that is 3 days late and 5% over budget today can become a 3-week delay and a $400K overrun by close — if no one flags the trend early enough.
The analytics engine (src/lib/analytics.ts) mirrors the logic you would want embedded in a Procore integration or any construction ops platform:
- Projects are automatically classified: On Time → At Risk → High Risk → Critical
- A risk score (0–100) is computed from three weighted signals:
- Schedule delay — up to 50 pts
- Budget overrun — up to 30 pts
- Delayed task count — up to 20 pts
- Alerts are auto-generated and severity-ranked across the full portfolio
- Insights are synthesized at the portfolio level (e.g., identifying that change orders are the primary cost driver across multiple at-risk projects)
| Metric | Purpose |
|---|---|
| On-Time Delivery % | Schedule health across all projects |
| Portfolio Budget vs Actual | Total cost exposure at a glance |
| Average Risk Score | Where to direct attention first |
| Active Alerts | What needs action today |
The Data Upload page accepts CSV — the export format Procore and most ERP systems support. Drop in a file and the entire dashboard rebuilds: KPIs, charts, alerts, and insights all update from the new dataset. This is the foundation for a real integration: pull from the source system, normalize, analyze.
The same architecture applied here to construction data generalizes directly to other operational domains:
| Domain | Data Source | Analytics Value |
|---|---|---|
| Construction | Procore | Schedule risk, cost forecasting, subcontractor performance |
| Human Resources | HRIS / ATS | Turnover signals, hiring funnel health, headcount planning |
| Logistics | ERP / TMS | Delivery delay patterns, vendor risk scoring |
| Layer | Technology |
|---|---|
| Framework | React 19 + TypeScript |
| Styling | Tailwind CSS v4 |
| Charts | Recharts |
| Build | Vite + Bun |
No backend required. Deployable as a static app and connectable to any REST API or CSV export from Procore, SAP, Workday, or a custom system.
# Install bun if needed
curl -fsSL https://bun.sh/install | bash
git clone https://github.com/incastil/construction-dashboard.git
cd construction-dashboard
bun install
bun run devOpen http://localhost:5173
src/
├── lib/
│ ├── analytics.ts # Risk scoring, alert generation, portfolio insights
│ └── formatters.ts # Currency, percent, date helpers
├── data/
│ └── mockProjects.ts # 8 realistic Procore-style projects
├── hooks/
│ ├── useProjects.ts # Central data, filter, and sort state
│ └── useTheme.ts # Dark mode with localStorage persistence
├── components/ # Cards, charts, table, alerts, insights
└── pages/ # Dashboard, Projects, Analytics, Upload, Settings
- Procore API integration — OAuth + REST to pull live project, budget, and RFI data
- LLM layer — Reason over unstructured change order descriptions and RFI notes
- HR module — Same architecture applied to headcount and attrition data
- Logistics module — Delivery performance and vendor risk scoring
- Role-based views — PM sees their projects; leadership sees the portfolio