Expert data and RL environments for AI labs.

Human judgment,
at lab quality
13,000+experts in our network
400+PhDs and MDs across six domains
6core domains

Expert judgment, turned into training data.

tasksrLdataAI LabsEXPERTSTASKSRL ENVIRONMENTSTRAINING DATAAI LABSEXPERTStasksrLRL ENVIRONMENTSdataAI LabsAI LABS

What We Build

Four things, done properly.

  • 01RL environments & agentic tasksTask environments with verifiers a model has to actually pass, built and stress-tested by people who do the work for a living.
  • 02Evaluations & rubricsBenchmarks, grading rubrics and model evaluations designed around how your team trains and tests, not a generic template.
  • 03Expert review & annotationDomain experts reviewing, ranking and correcting model output, with review built into the workflow you agree with us.
  • 04Multimodal & video dataCollection and review for video and multimodal datasets, from sourcing contributors to quality control.

Our Domains

Six domains. Deep, not wide.

Software engineering1,900+ computer science degrees
5,100+ experts
Machine learning & AI1,300+ with ML, AI or data science titles or degrees
3,200+ experts
Science & STEM560+ engineering, 290+ mathematics & statistics
1,800+ experts
Finance & accounting260+ finance or accounting degrees, 110+ economics
730+ experts
Medical & health70+ physicians (MD, MBBS or DO)
400+ experts
Legal65+ law degrees (JD, LLM or LLB)
130+ experts
Degree coverage and the work we staff in each domain →

How We Work

From first call to production, one team.

No vendor layers and no ticket queue. The people who scope your project are the people who build it.

01

Scope

You talk to the people who will build it. We agree the task spec, the quality bar and how we'll measure it.

02

Pilot

A small paid batch you can grade yourself. We fix what you flag before anything scales.

03

Scale

The same vetted experts, more of them, with review layers and regular reporting.

Trust

Your data stays yours.

  • 01Signed agreementsEvery contributor signs a contributor agreement, a confidentiality agreement and a code of conduct before seeing any project material.
  • 02Identity checksGovernment ID and selfie verification for the experts on your project when your work needs it, so you know who is doing it.
  • 03Confidential by defaultProject material we host is visible only to the experts on that project, behind their own sign-in.
  • 04A person answersExperts and clients reach a named person on our team, not a ticket queue.

Get Started

Tell us what you’re building.

A researcher on our team reads every message and replies personally.

For Labs

Work with us

Scope a pilot with the people who will build it.

Start a conversation ↗
Partner Work

ViaTouch × rustLabs

Training data for VICKI retail machines in the field.

See the case ↗
For Experts

Join the network

Paid, remote work in your field.

Go to rustLabs Experts ↗