Edge AI Digital Twin Developer

I’m building an on-premise platform that runs physics-informed digital twins on high-performance, specialized Edge AI computers already installed throughout a commercial building. The goal is simple: reduce energy use while keeping occupants comfortable, all in real time.

Here is what I need from you:

• A complete autonomous control loop that ingests live sensor data, runs the digital twin locally and pushes commands back to HVAC and lighting subsystems.
• Digital-twin models able to deliver all three key functions:
‑ Predictive maintenance
‑ Energy consumption analysis
‑ Environmental monitoring
• Code written primarily in Differential Swift. If a low-level module truly benefits from C, C++ or Python, I’m open to mixed-language solutions, but Swift should stay at the core.
• Seamless deployment to edge hardware: think small-form-factor GPUs, NPUs or other accelerator boards mounted in the electrical closet, not the cloud.
• Clear documentation and a repeatable build pipeline so future sites can be brought online with minimal effort.

I’ll consider the project complete once the system can run continuously on the building’s edge device for a full week, predict failures at least 24 h in advance, and show quantifiable energy savings versus the current baseline. If this sounds like the kind of challenge you thrive on, let’s talk details.