I’m building a carbon-aware orchestrator + digital twin for Indonesian AI data centers. Feedback welcome
Hi everyone, I’m Patrick, a Computer Science student in Semarang, Indonesia. I’ve been running my own online business since 2023, and lately I’ve been building EcoTwin AI.
The problem: AI data centers use a lot of power and water, and in Indonesia operators also have to deal with environmental permitting (AMDAL). Monitoring, workload scheduling and compliance reporting usually live in separate systems.
What EcoTwin AI does: it puts three things in one platform.
Carbon-aware orchestration: real-time inference gets priority, and batch training is deferred when grid carbon intensity is high.
Browser-based 3D digital twin: racks, thermal status and operational state, with no enterprise license needed.
Compliance metrics: PUE, WUE and CUE tracking with time-stamped, audit-friendly logs.
It’s tuned for Indonesian conditions, such as geothermal baseload and variable tropical solar.
Where it stands today: the core engine (carbon analyzer, orchestrator, metrics, compliance logger) works, and there’s a 3D dashboard prototype you can open in a browser with no setup. The grid and sensor data is still simulated. Next on the roadmap are live PLN/renewable-energy data, real facility sensors, and a multi-tenant backend.
Where I’d love feedback:
Is carbon-aware scheduling realistic for data centers in Indonesia, given how the grid is structured? If you’ve worked with data center monitoring, which metrics beyond PUE/WUE/CUE matter most? What’s the best way to get real grid data in Indonesia?
0 comments