Field notes

From the data lab.

Essays and technical notes on producing real-world data for physical AI: the datasets, the methods, and the systems we build.

How-ToJul 14, 20264 min read
How to run your first data-vendor pilot without regressing model quality
A repeatable playbook for testing an AI data vendor on a small, real slice of work and proving quality on a held-out eval before you move volume.
Mohtada Jokhio
MigrationJul 13, 20265 min read
Migrating from Scale AI to Synthsis Labs
A practical guide to migrating from Scale AI to Synthsis Labs: a more cost-effective, tailored data partner for mid-sized labs, plus a step-by-step switch plan.
Rasikh Ali
AlternativesJul 13, 20264 min read
Scale AI Alternatives for Mid-Sized AI Labs
A practical look at Scale AI alternatives for mid-sized labs, including Synthsis Labs, Surge AI, Mercor, and Turing.
Armaghan Ejaz
ComparisonJul 13, 20265 min read
Synthetic vs. real-world data for robotics: what actually trains a policy
Simulation is cheap and infinite; real-world data is expensive and finite. Here is how the two actually compare for robotic manipulation, and why the strongest training sets use both.
Mohtada Jokhio