INeed to have solution (like 2wai) for to have full avatar in real time , keeping realistic approch etc. Technology should be gaussian splatting or different one allowint got create something like this. On top voice cloning is needed, User should be able ot generate it on website and use it later on on external devices am building a web-based creator that lets anyone generate a full-body avatar—similar in quality to 2wai—The result must look highly realistic, respond instantly, and be usable later on external devices (mobile, led walls) We should have option to talk in offline mode Core requirements • A capture-to-avatar pipeline that can rely on Gaussian Splatting, NeRF, or any other method you propose • Real-time facial expressions, body movements and perfect lip-sync driven by either live video or prerecorded motion data. • Integrated voice cloning so the avatar speaks with the user’s cloned voice on the fly. Acceptance criteria 1. A working web demo running on commodity hardware with <120 ms end-to-end latency. 2. Clear build / deploy docs plus commented source or compiled modules. 3. An API or SDK sample showing the avatar streamed into a third-party Unity or Unreal scene. If you have shipped anything comparable—volumetric capture, NeRF avatars, voice cloning pipelines—tell me which parts you can deliver fast and where you would need collaboration. I am ready to iterate quickly and can supply test footage, hosting and compute. IP agreement will be required