← David Díaz Tunjano

MELO CREO — AI career guidance for public schools

Role: Lead AI Software Engineer

MELO CREO fights an invisible problem: in Medellín, thousands of students finish high school without a clear path — not for lack of scholarships, but for lack of information at the right moment. I built the entire product technology end to end. The system combines a diagnostic survey, an interactive learning platform, and three conversational AI agents — a Vocational Bot, the MELO Agent, and a Scholarship Advisor — that answer each student's specific questions about careers, admission requirements and financing. Built with academic-grade evaluation (an RCT is underway for 2026) — impact measured with causal humility, not marketing claims.

  • 102 → 131
    Public schools (2025 → 2026)
  • 13,687
    Students in the 2026 diagnostic
  • 23,027
    Messages across bot sessions
  • +22.6%
    "Exploring" → "decided"
  • 13.2%
    ITM enrollment vs. control (7.4%)
  • 92.4%
    Grade-promotion (high engagement)
  • Python
  • LangGraph
  • LLM Agents
  • n8n
  • AWS (EC2, Lambda)
  • Docker
  • FastAPI
  • React
David Díaz presenting MELO CREO to students in a computer lab in Medellín
Rolling out MELO CREO in the field — La Avanzada school, Medellín.
Students completing the MELO CREO diagnostic on laptops
13,687 students started the 2026 diagnostic. This is what that looks like.
A full computer lab running MELO CREO, with the MELO robot mascot on screen
A whole class on MELO at once — Villa del Socorro, Medellín.
“I really liked talking to the guide — it helped me figure out which career I wanted. MELO's AI is great.”
— Samuel García, 11th grade, Tricentenaria School (Medellín)

Visit live site →