Wildflow uses Nebius AI Cloud, including Serverless Jobs, GPUs and shared persistent storage, to turn terabytes of underwater footage, acoustics and environmental data into high-fidelity 3D coral reef models. Its reconstruction pipeline works with data from more than 20 research organizations worldwide. Researchers can explore the web-based digital twins and compare reef conditions over time to support monitoring and restoration. Aleksandr Patrushev describes potential expansion from dozens of reef sites to hundreds worldwide.
The pipeline combines GPU-intensive training, chunking and tiling with CPU-bound camera positioning. Its custom 3D Gaussian Splatting architecture produces precisely georeferenced digital twins. Nebius shared filesystems retain full-resolution imagery through preprocessing, alignment, reconstruction and rendering. Serverless Jobs scale to more than 20 parallel jobs, each using its own container image, with agents helping automate the workflow. On-demand HPC virtual machines and shared filesystems support the pipeline. A full reef reconstruction takes under a day, and 3D Gaussian Splatting rendering takes a few hours.
Wildflow aims to train multimodal foundation models for biodiversity research that interpret sensor, acoustics, DNA and biogeochemistry data. Its digital twins provide a foundation for modeling predator-prey dynamics, population growth and coral spawning patterns to inform restoration. It also aims to extend its ecosystem modeling framework to all major Earth biomes within eight years.
Nebius describes Wildflow using Nebius AI Cloud, Serverless Jobs and shared filesystems for coral reef reconstruction. Wildflow aims to develop foundation models interpreting sensor, acoustics, DNA and biogeochemistry data. Its digital twins provide a foundation for modeling predator-prey dynamics, population growth and coral spawning patterns to inform restoration.
Nebius Customer story24 Sept 2026 · Nebius · Web pageNebius describes Wildflow using Nebius AI Cloud, Serverless Jobs and shared filesystems for coral reef reconstruction. Wildflow aims to develop foundation models interpreting sensor, acoustics, DNA and biogeochemistry data. Its digital twins provide a foundation for modeling predator-prey dynamics, population growth and coral spawning patterns to inform restoration.
Describes Wildflow's reef-reconstruction pipeline using GPU jobs and shared storage
31 Aug 2026 · Avpatrushev (LinkedIn) · Web page
Aleksandr Patrushev describes Wildflow running its reef reconstruction pipeline on Nebius using Serverless Jobs, GPUs and shared filesystems. Jobs scale to more than 20 in parallel. Full reef reconstruction takes under a day, and 3D Gaussian Splatting rendering takes a few hours. He quotes founder Sergei Nozdrenkov describing on-demand Nebius HPC virtual machines and fast persistent storage.
Describes Wildflow's reef-reconstruction pipeline using GPU jobs and shared storage31 Aug 2026 · Avpatrushev (LinkedIn) · Web pageAleksandr Patrushev describes Wildflow running its reef reconstruction pipeline on Nebius using Serverless Jobs, GPUs and shared filesystems. Jobs scale to more than 20 in parallel. Full reef reconstruction takes under a day, and 3D Gaussian Splatting rendering takes a few hours. He quotes founder Sergei Nozdrenkov describing on-demand Nebius HPC virtual machines and fast persistent storage.
Wildflow develops biodiversity software focused on coral reef monitoring and restoration. Its Wildflow Coral product turns underwater photographs and survey footage into browser-based 3D reef models that researchers and restoration teams can explore and share. The platform supports reef mapping and analysis of coral growth, survival, habitat structure and other health indicators. Wildflow is also developing multimodal AI models that combine imagery with acoustic, environmental and DNA data to help scientists understand ecosystem change and plan conservation interventions.
Wildflow uses Nebius AI Cloud, including Serverless Jobs, GPUs and shared persistent storage, to turn terabytes of underwater footage, acoustics and environmental data into high-fidelity 3D coral reef models. Its reconstruction pipeline works with data from more than 20 research organizations worldwide. Researchers can explore the web-based digital twins and compare reef conditions over time to support monitoring and restoration. Aleksandr Patrushev describes potential expansion from dozens of reef sites to hundreds worldwide.
The pipeline combines GPU-intensive training, chunking and tiling with CPU-bound camera positioning. Its custom 3D Gaussian Splatting architecture produces precisely georeferenced digital twins. Nebius shared filesystems retain full-resolution imagery through preprocessing, alignment, reconstruction and rendering. Serverless Jobs scale to more than 20 parallel jobs, each using its own container image, with agents helping automate the workflow. On-demand HPC virtual machines and shared filesystems support the pipeline. A full reef reconstruction takes under a day, and 3D Gaussian Splatting rendering takes a few hours.
Wildflow aims to train multimodal foundation models for biodiversity research that interpret sensor, acoustics, DNA and biogeochemistry data. Its digital twins provide a foundation for modeling predator-prey dynamics, population growth and coral spawning patterns to inform restoration. It also aims to extend its ecosystem modeling framework to all major Earth biomes within eight years.
Nebius describes Wildflow using Nebius AI Cloud, Serverless Jobs and shared filesystems for coral reef reconstruction. Wildflow aims to develop foundation models interpreting sensor, acoustics, DNA and biogeochemistry data. Its digital twins provide a foundation for modeling predator-prey dynamics, population growth and coral spawning patterns to inform restoration.
Nebius Customer story24 Sept 2026 · Nebius · Web pageNebius describes Wildflow using Nebius AI Cloud, Serverless Jobs and shared filesystems for coral reef reconstruction. Wildflow aims to develop foundation models interpreting sensor, acoustics, DNA and biogeochemistry data. Its digital twins provide a foundation for modeling predator-prey dynamics, population growth and coral spawning patterns to inform restoration.
Describes Wildflow's reef-reconstruction pipeline using GPU jobs and shared storage
31 Aug 2026 · Avpatrushev (LinkedIn) · Web page
Aleksandr Patrushev describes Wildflow running its reef reconstruction pipeline on Nebius using Serverless Jobs, GPUs and shared filesystems. Jobs scale to more than 20 in parallel. Full reef reconstruction takes under a day, and 3D Gaussian Splatting rendering takes a few hours. He quotes founder Sergei Nozdrenkov describing on-demand Nebius HPC virtual machines and fast persistent storage.
Describes Wildflow's reef-reconstruction pipeline using GPU jobs and shared storage31 Aug 2026 · Avpatrushev (LinkedIn) · Web pageAleksandr Patrushev describes Wildflow running its reef reconstruction pipeline on Nebius using Serverless Jobs, GPUs and shared filesystems. Jobs scale to more than 20 in parallel. Full reef reconstruction takes under a day, and 3D Gaussian Splatting rendering takes a few hours. He quotes founder Sergei Nozdrenkov describing on-demand Nebius HPC virtual machines and fast persistent storage.
Wildflow develops biodiversity software focused on coral reef monitoring and restoration. Its Wildflow Coral product turns underwater photographs and survey footage into browser-based 3D reef models that researchers and restoration teams can explore and share. The platform supports reef mapping and analysis of coral growth, survival, habitat structure and other health indicators. Wildflow is also developing multimodal AI models that combine imagery with acoustic, environmental and DNA data to help scientists understand ecosystem change and plan conservation interventions.
