Make the right robotics data bet
Work directly with Sam Padilla on the hard decisions behind robotics data, collection hardware, evaluation, and deployment. Get an experienced outside view before you spend months scaling the wrong approach.
Built from experience
Sam co-founded Eidon AI and worked across the full robotics data stack, from wearable capture hardware and mobile apps to simulation, quality control, global collection, and delivery to frontier robotics labs.
Where I can help
The best advisory work starts with a consequential decision, a real constraint, and enough context to challenge the obvious answer.
Market and customer strategy
Pressure-test the buyer, deliverable, defensibility, and business model before scaling a collection program or product.
Capture and hardware
Choose modalities, sensors, embodiment assumptions, and build-versus-buy boundaries around the model objective.
Collection operations and QA
Design field operations, contributor workflows, calibration, quality gates, and delivery systems that hold up outside a demo.
Evaluation and deployment
Connect data decisions to measurable model performance, failure analysis, simulation, and reliable behavior in the real world.
How the work runs
Come with a decision. Leave with a sharper one, the tradeoffs behind it, and the next actions to take.
Context
Share the relevant architecture, customer signal, research constraints, and decisions before the call.
Working session
We dig into the problem, challenge assumptions, compare options, and make the tradeoffs explicit.
Decision memo
You receive a concise written recap with recommendations, risks, and concrete next steps.
Follow-through
Retainer clients use the next calls and async support to pressure-test what changes as the work moves.
Advisory options
Start with one working session. Use the retainer when the decisions keep moving and the cost of a wrong turn is high.
A focused session for a specific technical, product, or market decision.
- 75-minute working session
- Materials reviewed in advance
- Written recommendations and next steps
Ongoing decision support for founders and teams actively building in robotics or physical AI.
- Two 60-minute calls each month
- Limited async questions and document review
- Three-month minimum
Implementation, embedded operating work, and recurring team meetings are scoped separately.
Questions
Who is this for?
Founders and technical teams working on robotics, physical AI, data collection, simulation, evaluation, or deployment. It is especially useful before a major hardware, hiring, or collection investment.
What should I bring to a strategy session?
Send the product context, current architecture, open decisions, and any customer or research constraints ahead of time. The more specific the decision, the more useful the session will be.
What does async support include?
Short questions, document feedback, and decision support between scheduled calls. It does not include implementation, recurring team meetings, or embedded operating work.
Can you help execute the recommendations?
Yes. Architecture, prototyping, data systems, collection tooling, and other hands-on work can be scoped separately through By Crux.
Can you sign an NDA?
Yes. Confidential technical and commercial details are common in this work, and an NDA can be signed before materials are shared.
Make the expensive decisions once
Bring the problem, the constraints, and the uncomfortable questions.