Procergs | School Drop Out Prevention

AI-powered prediction
Early intervention for educators

Design Leadership, Information Architecture, and User Experience

Situation

In 2022, Rio Grande do Sul faced a education crisis: an 11% school dropout rate — nearly double Brazil's national average of 6.6%. While several social policies existed to address the issue, Procergs (the state's IT agency) recognized that with vast amounts of data available, they could take a proactive, data-driven approach.

Task

Lead the design of an AI-powered early intervention system that would help schools identify and support students at risk of dropping out, focusing on making complex AI predictions actionable for busy school administrators.

Design Challenge:
Create an experience that would help Pedagogical Coordinators act on AI predictions without adding complexity to their workflow.

Actions (1 to 4)

1. Stakeholder Research
Conducted individual interviews across Procergs and Government teams to map expectations, and information priorities before the collaborative workshop.

 

2. Workshop & Persona Mapping
Facilitated in-person workshop mapping all personas in the dropout prevention ecosystem and identifying the Pedagogical Coordinator as our primary user — the person who could act on AI predictions.

 

3. Available & Feasible Data
Collaborated with the technical team to map available student data and sources.

Solutioning workshop in person

4. Key Design Decision: Integration Over Addition

  • Technical team's proposal: Build WhatsApp assistant to message Pedagogical Coordinators with at-risk student alerts.
  • My research insight: Pedagogical Coordinators were already overwhelmed with WhatsApp messages. Adding another notification channel would be ignored.
  • Design solution: Integrate AI predictions into the existing system they used daily, the infrequent student attendance list.

Why this worked:
Designed a interface that prioritized simplicity and action.

  • No new tool to learn or check
  • Students ranked by dropout probability
  • Notes system for Coordinators to document conversations with students, families, and teachers
High fidelity of the List and Actions for the Pedagogical Coordinator

Results

Immediate Impact

  • Predictive model with over 91% accuracy
  • Procergs and RCI teams moved quickly to production deployment

Long-Term Success (2024)

  • AI model validated in four schools with satisfactory results
  • Demonstrated IBM's capability in civic tech and social impact AI

Selected work

Let's connect

That’s the end of the scroll, but just the start of the conversation.
I’m currently open to new collaborations and design leadership roles.