Notes from the certification frontier.
Engineering deep dives, research, and product updates from the team building the readiness layer for Physical AI.
Calibrating the sim-to-real gap estimator
How we turn twin-vs-reality deltas into a defensible readiness score.
Jonas Meyer · 11 min
Running 50k parallel trials without melting your budget
Elastic GPU scheduling for large validation sweeps.
Devon Aku · 7 min
A shared language for robot failure modes
Toward a taxonomy safety and ML teams can both use.
Priya Varma · 9 min
Introducing the Counterfactual Scenario Miner
Find the edge cases that actually change a verdict.
Tomas Kite · 6 min
OpenUSD twins that match real hardware
Calibration techniques for contact-rich manipulation.
Ada Reyes · 10 min
Why we're building the certification layer
The founding thesis behind Certisto.
Ada Reyes · 5 min
World models for the long tail
Where classical randomization stops and synthesis begins.
Jonas Meyer · 12 min
Gating deployments on readiness in CI
Make a minimum certified score part of every release.
Devon Aku · 6 min
What a certified operating envelope means
The conditions a policy is actually cleared to run in.
Priya Varma · 8 min
Deep dives by theme.
How we write.
- Show the method, not just the result
- Report the trade-offs and what didn't work
- Prefer real numbers over adjectives
- Written by the people who built it
Honest by default
We publish limitations alongside wins.
Reproducible
Methods detailed enough to follow along.
Popular topics.
From the engineering blog
Written by the people building it.
Our researchers and engineers publish the methods, trade-offs, and lessons behind the platform — no marketing gloss.
Ada Reyes
Jonas Meyer
Priya Varma
Devon Aku
New posts in your inbox.
One email a month with our best engineering and research writing.
Prefer to see it live?
Skip the reading list — get a certificate on one of your own policies.