Back to selected work

Focus · Flagship product

From school incidents to timely intervention

Focus helps school teams record incidents, spot patterns, manage follow-up and ask questions in Portuguese. It brings the web app, PostgreSQL data, dashboards, alerts and AI into one workflow.

AnalyticsData EngineeringApplied AI
Case study available
4role-specific experiences
2AI providers with fallback
PDF + XLSXoperational exports
E2Econfigured browser tests

Product presentation

See Focus in action

A short presentation of the Focus app, from recording school occurrences to finding patterns and following up.

The problem

What needed to change

A list of incidents is not enough for a school team. Staff need to record what happened, protect each school’s information, notice patterns, contact the right people and follow each case until it is resolved.

My approach

How I approached it

I designed Focus as a complete working tool rather than a standalone dashboard. Data entry, access rules, analysis, alerts, reports and the AI assistant all support the same day-to-day process.

System flow

How the pieces connect

  1. 01Role-aware capture
  2. 02Next.js API layer
  3. 03Supabase + PostgreSQL RLS
  4. 04ECharts analytical views
  5. 05Gemini / Groq AI layer
  6. 06Alerts, reports, and follow-up

What I built

What is included

  • School-aware access rules built with PostgreSQL Row Level Security.
  • Master, administrator, professor, and read-only viewer workflows.
  • Student, class, occurrence, feedback, alert, and academic-year management.
  • Interactive charts that let users explore related patterns across the dashboard.
  • Natural-language questions over school data with Gemini and Groq fallback.
  • Excel/PDF exports, email notifications, imports, and mobile navigation.

Key decisions

Choices that shaped the project

01

Keep each school’s data separate

Row Level Security supports school-aware access in the standard data paths. I am keeping the public demo closed until the service-role routes complete an additional authorization review.

02

Keep the assistant available

The assistant can use a second model provider when the primary provider is unavailable or reaches a usage limit.

03

Connect insight to follow-up

Charts, alerts, status changes, feedback history and exports live in the same product so the team can act on what it finds.

Result

What I delivered

Focus shows how I work across product design, data modeling, analysis and applied AI. It takes a real operational task from data entry through follow-up.

Next improvement

What I would do next

My next step is to finish the security review, create a safe demo school with sample data and record a short walkthrough for recruiters.

Tools and methods

Next.jsTypeScriptPostgreSQLSupabaseEChartsGeminiGroqPlaywright

Next case study

Job Hunter Agent

A Python workflow that gathers remote data jobs, removes duplicates, evaluates fit with several AI providers, prepares tailored documents and sends the strongest matches to Telegram.

Continue exploring