Research Frameworks
Shared instruments for reproducible AI science.
Research Frameworks are the internal and external toolkits that make Zansoc labs productive: experiment tracking, dataset contracts, eval standards, and simulation environments. Six laboratories share one instrument set so results compose instead of fragmenting.
Experiment OS
Reproducible runs with full artifact capture.
Eval Standards
Shared metrics across labs and products.
Simulation
Worlds for robotics, agents, and cities.
Publication Tooling
From notebook to paper-ready figures.
One instrument set
Vision Lab detections can feed Robotics Lab planners because schemas and eval harnesses agree. Research Frameworks enforce those contracts without slowing iteration.
From lab to product
When an experiment graduates, the same eval definitions travel into product CI. Customers inherit the measurement culture, not just the weights.
Open where it helps
Selected frameworks and benchmarks are published alongside papers so the broader community can reproduce and extend our results.