Service I
Contract research
A working, verified model that answers the client's question about its own system.
- The outcome
- The client brings a question about a changing system and receives an executable answer: a causal model built on MIT-licensed DeepCausality crates, and a report stating what the model shows and where its claims stop. Typical questions ask which action keeps a vehicle within limits after a fault, where a network failure began, or what a flow would have done under a different decision.
- The evidence
- The Center publishes its methods and code, so the client can inspect the foundation before commissioning work. The Center's own results set the standard for delivered work. A chronometric inversion recovered Earth's gravitational parameter from Galileo satellite clock data to a median precision of 4 × 10−7. A drone fail-safe cut landings near people from 3.5 % to 0.1 % across 1,000 simulated emergencies. Counterfactual Fluid Dynamics checks its results in thirteen verification programs.
- Time to first result
- Work starts from a working stack, not a blank page: DeepCausality already provides causal discovery, composition, safety norms, and physics and fluid-dynamics solvers. A six-to-ten-week first phase scopes the question and delivers a working model. Each later phase begins only if the previous result justifies it.
- How to start?
- Contract research suits an organization with a specific question about a system that changes while it runs. It requires data or a system description and a domain expert for a few hours a week. An exploratory call identifies the question and scopes the engagement.
Starting an engagement
The client describes the problem in a few lines, and the Center replies to arrange the first call. Client data and client-specific results stay with the client; verified general methods return to the open DeepCausality project.