Multi-agent systems for autonomous driving at Uber
Consulting on how fleets of models coordinate for autonomous driving: agent communication, simulation-first evaluation and the verification gates between a model's decision and the road.
Sector: Mobility · Uber · details anonymized
The challenge
Autonomous driving is a multi-agent problem twice over: perception, prediction and planning models must agree with each other inside one vehicle, and the development organization needs thousands of scenario evaluations to trust any change.
The coordination and evaluation machinery around the models, not the individual models, is where reliability is won or lost.
Our approach
Mindela's founder consulted on agent coordination and evaluation architecture for the autonomous-driving effort: structured communication between agents, simulation-based scenario suites as the gate for changes, and verification steps between model output and actuation.
Architecture
Simulation-first evaluation: no change ships on vibes; scenario suites decide.
Structured agent-to-agent contracts instead of implicit coupling between models.
Outcome
Engagement delivered by Mindela's founder as a consultant. Specifics are shared in calls.
Building something similar?
We're happy to talk through how this architecture would map to your problem. No pitch, just engineering.
Start the conversationMore work
An AI assistant that resolves queries end-to-end
A tool-using AI assistant grounded in the client's own knowledge. It answers, acts, and escalates to humans when confidence drops.
An agent harness for long-running autonomous work
Claude-Code-style orchestration with planning, tool execution, retries and evals, so an LLM can carry multi-step tasks to completion without babysitting.