Physical AI · Demonstration Data

We train the
motion behind
the machines

Humanoids don't learn from manuals — they learn from human demonstrations. Nostromo Labs collects, annotates, and packages that data for embodied-AI teams: egocentric video, episode-level labels, and teleoperation-ready pipelines.

TZ UTC−6 FIRST BATCH 48H QA PER-EPISODE
rig/humanoid_v2 · kinematic chain capturing
shoulder_pitch41.2°
elbow_flex88.7°
wrist_roll−12.4°
grip_force6.3 N
episodeEP-0147
label✓ success

Capabilities

The data layer for embodied AI

Robot foundation models are bottlenecked by manipulation data. We run the human side of that pipeline so your researchers stay on the model side.

Collection

Egocentric demonstrations

First-person video of real hands performing real tasks — kitchens, workshops, assembly lines — captured by our trained contributor network across the Americas, following your task scripts and camera protocols.

ONLINE — available now

Annotation

Episode labeling & QA

Success / failure / recovery tags, grasp events, object tracks, and language instructions on your teleop or video data. Cleaned, deduplicated, and packaged for training — with a documented QA trail per episode.

ONLINE — available now

Teleoperation

Managed teleop cell

Certified operators generating demonstrations on client-matched hardware — scenario scripting, in-session QC, and delivery straight into your training pipeline, from our nearshore facility.

WAITLIST — pilot partners

Coverage

Task domains & data modalities

We speak your taxonomy. Specialized in deformable-object and household manipulation — the hardest data to source — with full coverage across domains and modalities.

Task domains

Pick & place Deformables — cloth, cables Articulated — doors, drawers Kitchen & household Industrial assembly Tool use Bimanual tasks Pouring & liquids Human–robot handover

Data modalities

Egocentric RGB video Multi-view video Speech & voice commands — EN / ES accents Audio events Language instructions Success / failure / recovery labels Temporal segmentation Grasp events & keypoints

Standing catalog. A growing library of pre-collected, licensable episode sets — Latin American homes, workshops, and voices that public datasets don't cover. Ask for the current index.

Request index

Pipeline

Spec to training-ready

T+0

Task spec

You define tasks, environments, and success criteria. We convert them into collection scripts and operator briefs.

T+48h

Pilot batch

A labeled sample in your exact format, so you validate framing and label quality before scaling.

SCALE

Parallel capture

Contributor network and capture team run concurrently with in-session QC and daily throughput reports.

SHIP

Delivery + QA trail

Datasets streamed into your training pipeline with per-episode review logs.

Operations

Nearshore by design

Built to work the way US robotics programs actually procure: an American entity to contract with, and a nearshore operation you can visit in a four-hour flight.

OPS San José, Costa Rica · lat 9.93 lon −84.08 ENTITY US LLC for contracting & invoicing NETWORK Contributors across Latin America
Your working hours are our working hours UTC−6 year-round: standups, live QC sessions, and low-latency remote teleop without midnight handoffs.
A workforce built by precision manufacturing The talent pool behind decades of medical-device and electronics assembly — people who already think in protocols, tolerances, and traceability.
Bilingual by default English-fluent project leads, native-Spanish contributor base: one partner for English and Spanish instruction data.
Hardware-ready logistics Free-trade-zone import channels and direct flights to every US hub for client-matched teleop hardware.

Start small · Ship fast

Get a pilot dataset in your format this week

Tell us one task your robot struggles with. We'll return a labeled sample batch — egocentric video, audio, or annotation on your data — so you judge quality before committing to anything.

Response within one business day · UTC−6

or write us directly — lester@nostromolabs.lat