topological data analysis

Axion Deep Labs is an independent R&D company that publishes original research in machine learning, quantum computing, and topology, then builds it into products and licensed IP. It runs experimental programs in deep learning theory, including continual learning, topological data analysis, information capacity, and loss landscape geometry, with reproducible, open-source methodology.

ClearML experiment managementdeep learning researchopen source software developmentPyTorch model trainingquantum computing researchreproducible experimentsRipser/scikit-tda topological computation

BluelightAI provides AI interpretability for systems in production. It reads existing test and production traces at the concept level to surface behavior that evaluations never described, showing why a model failed before it reaches production. It serves data science, risk, and engineering teams in banking, financial services, insurance, healthcare, and defense, using Topological Data Analysis and mechanistic interpretability.

coverage gap analysiscross-layer transcodersMechanistic Interpretabilitysparse autoencoderstrace analysis

Geometric Data Analytics (GDA) is a research, development, and consulting company focused on solving complex data analysis problems. Its team includes key founders of topological data analysis and experts in differential equations, machine learning and AI, physics, biology, applied mathematics, data visualization, and computer science. GDA develops agile, test-driven software and works with government agencies, university labs, and private research teams.

CI/CD pipelinescontainerized microservicesdata visualizationdifferential equationsimage analysismachine learning and artificial intelligencesoftware engineeringTest-Driven Developmentverification and validation of AI

Krv Labs builds evaluation infrastructure for agentic AI systems. It produces evidence packages that justify an agent's decisions and expose uncertainty by inspecting the data (Pulsar), models (Pasteur), and code (Topos) behind each decision, rather than just inputs and outputs. Its core verification libraries are open source under the BSD 3-Clause License.

AI agent evaluationAI-generated code auditingdata manifold analysismodel stress-testingopen source softwarerepresentation learning