EcoLafaek
The crocodile that guards Timor-Leste
An AI-powered waste management system connecting citizens, authorities and technology to fight the waste crisis in Timor-Leste.
Díli produces over 300 tons of waste daily, with more than 100 tons left uncollected. The waste clogs drains, causes floods and damages health and the environment.
Our vision is a cleaner Timor-Leste where communities can report environmental issues, authorities can prioritise cleanup, and environmental health keeps improving.
Citizens use the Flutter mobile app to report waste with a photo and location. It works in low-connectivity areas.
Amazon Nova-Pro classifies the waste type, assesses severity and estimates the environmental impact of every photo.
The data puts cleanup where it matters most. Municipal teams get the jobs, and every report is tracked to resolution.
Interactive maps, hotspots, analytics and leaderboards, open to everyone, with an AI agent that answers questions.
Built with AWS AI services and modern tools for an autonomous, scalable waste management system.
Image analysis, waste classification and autonomous reasoning with multi-round tool calling.
Code Interpreter and Browser tools for SQL queries, chart generation, map creation and web research.
amazon.titan-embed-image-v1 for semantic search and similarity matching with 1024-dimension vectors.
Citizen reporting app with GPS, offline support and photo capture; Staff app for cleanup teams.
Web applications with API routes, real-time analytics and interactive maps.
Pydantic validation, JWT authentication and AgentCore orchestration.
Serverless routes with direct database access.
MySQL-compatible database with vector search and automatic scaling.
Ubuntu servers with SSL and managed services for the backend.
Report photos, generated charts and static assets.
Global deployment for the Next.js applications.
Spec-driven development with autonomous agents.
Multi-round tool calling on the AgentCore runtime
Semantic similarity with Titan embeddings
Live dashboards with interactive visualisations
JWT auth, rate limiting and encrypted data
EcoLafaek is an AI-powered system that addresses waste management challenges in Timor-Leste through a network of intelligent agents. It enables citizens to report waste issues via a mobile app, provides authorities with actionable insights through an analytical dashboard, and supports data-driven decisions for waste management.
Download the mobile app to report waste in your area, explore the public dashboard to see waste hotspots, or request access to our code repository to collaborate.
Amazon Nova-Pro's multimodal capabilities analyse waste images, classify waste types, assess severity and identify potential environmental impacts. The system also detects waste hotspots with spatial clustering.
Report locations and images are shared publicly to help address waste issues, but account security uses JWT tokens with PBKDF2-SHA256 password hashing, and personal data is never shown publicly.
Lafaek means “crocodile” in Tetum: a sacred guardian in Timorese legend, a symbol of our commitment to protecting Timor-Leste's environment.
Download the app, report waste incidents and become part of the solution for a cleaner Timor-Leste.