From the streets of Díli to the biggest stage in AI, EcoLafaek never stopped showing what a small country can build.
A journey made of photos, code and community. This is the story of a Timorese project, told through the places that shaped it: the streets where it was born, the stages that crowned it, and the people at home who celebrated it.

Díli produces more than 300 tons of waste every day, and over 100 tons of it are never collected.
It blocks the drains, floods the streets and ends up in the rivers and the sea. Nobody had a map of where waste builds up, and citizens had no simple way to report it. EcoLafaek started here, in the neighbourhoods of Díli.




A citizen takes a photo of waste. Amazon Nova-Pro reads it: the type, how severe it is, how much there is.
Clear reports verify themselves. Municipal teams receive cleanup jobs in the Staff app, and everyone follows the progress on the public dashboard, where an AI agent built on Amazon Bedrock AgentCore answers questions with charts and maps.

Fetch.ai Global AI Agents League 2025: third place and the AI Agent Trailblazer Prize, out of 0 participants, for the TL Digital Waste Monitoring Network. The first Timorese to win an international AI competition.
On the re:Invent keynote stage, Dr. Swami Sivasubramanian, VP of AI & Data at AWS, announced the first place winner in front of the world: EcoLafaek, built in Timor-Leste.




“For redefining what’s possible in grassroots waste management with a creative agentic solution built with Amazon Bedrock Nova-Pro powered by AgentCore to improve his city.”

In 2026 the EcoLafaek story travelled with AWS: featured in Summit keynotes across eight world cities.









“Parabéns ba jovem Timor oan ida né”
Congratulations to this young son of Timor.

Selected among the 24 finalist teams of the Korea-ASEAN AI Youth Festa 2026. Next stop: pitching in Manila, 3 to 6 November 2026, with Timor-Leste on the map.
A community map of every public rubbish bin in Timor-Leste.
Municipal cleanup teams get jobs, routes and before/after photos.
Ask the data anything: charts and maps answered by Bedrock AgentCore.