Core Frameworks
PyTorch Geometric
GNN and geometric deep learning — our primary research and production framework.
JAX & Flax
Differentiable programming and hardware-accelerated manifold computations.
Giotto-TDA
Topological data analysis pipelines fully integrated with scikit-learn.
Hugging Face
Transformers, diffusers and the full LLM ecosystem for language tasks.
Ray & Dask
Distributed computing for large-scale manifold training and inference.
MLflow / W&B
Experiment tracking with geometric metric logging and model registry.
Kubernetes
Container orchestration for portable geometric AI deployment anywhere.
AWS/GCP/Azure
Multi-cloud strategies with GPU cluster management for large-scale training.
The Flux Pipeline
01Data Topology Analysis
02Manifold Feature Engineering
03Geometric Model Training
04Topological Evaluation
05Continuous Deployment
06Flux Monitoring & Retraining