Research becomes guidelines.
We take frontier research and turn it into guidelines.
Skip to content
01 / Build 02 / Gate 03 / Search 04 / Grade 05 / Comprehensive 06 / Place
01 / Learn 02 / Practice 03 / Improve 04 / Qualify 05 / Keep it 06 / Stay connected
The things people
For post-training teams · Book a walkthrough
High quality datacomes from highlytrained evaluators
Data Gradient trains annotators to produce high quality data your models need now.
Staffing searchPlanned for teams
Search qualified people instead of resumes
Teams will be able to search the qualification record by task type, language and modality instead of interviewing their way through a spreadsheet of resumes.
FIND
A. Rivera
ES · MX · 14 MO ACTIVE
94.2
Verified 3d ago
M. Okonkwo
ES/EN · ES · 9 MO ACTIVE
91.8
Verified 1w ago
J. Bautista
ES · PH · 22 MO ACTIVE
89.5
Verified 2d ago
L. Fernández
ES · AR · 5 MO ACTIVE
86.1
Re-drill due
Scores come from graded practice on your rubric. Self-reported experience doesn't count. Every one is re-verified on a schedule you set.
01 The ideaN 01 / 06
We don’t want to train models. We want to train people.
Human data companies and AI labs work rigorously to post-train, fine-tune, and evaluate frontier models. Data Gradient applies that same level of rigor to training the workforce.
Model sideData Gradient
Mid-trainingFoundational courses built from guidelines on domain-specific corpora
Supervised fine-tuningGuided drills on ideal response generation that improve outcomes
EvalsScored practice against production rubrics to inform your next training run
RLHFReviewer feedback loops on real output with reasoning traces
Continual trainingSpaced re-trains to maintain high-quality outputs for persistent projects
02 Two tracks, one recordAnnotators → teams
Annotators build their credentials. Teams hire it.
The same qualification data serves both sides. Annotators earn a verified, portable record of what they can actually do. Teams will be able to search that record instead of interviewing their way through a spreadsheet of resumes.
Annotator track
↓- 01 Learn the workflow
- 02 Drill in a sandbox
- 03 Get graded
- 04 Cross the gate
Verified record
↓- Skills · scores · error classes
- Task types · languages
- Freshness · decay
- Owned by the annotator
Team track · planned
- 05 Search the pool
- 06 Staff the project
- 07 Monitor quality live
- 08 Re-drill on drift
Production quality feeds back into the record
03 For teamsSix reasons
Qualifications you can count on.
Built for the people accountable for batch acceptance: technical program managers, strategic project leads, delivery and quality leads.
Credentials that matter.
If an annotator doesn't pass the threshold, they don't get a certificate.
We have the metadata.
We track everyone's ability by task type, language, and modality.
Rubrics mean something here.
They mean everything. We take rubrics very seriously. Every task is answered in accordance with a clearly outlined rubric.
One program, many techniques
We plan to cover every modality known to man... in due time.
Partner and receive
If you partner with us on a course, you get vetted candidates from that course.
04 For annotatorsSix reasons
Training that labs actually need
We take state-of-the-art research and create training courses out of it. Learn also lists more than 1,200 jobs from human data companies and marketplaces, refreshed regularly.
Courses for real work
We run our courses the way we ran our teams for the biggest frontier labs on the planet.
Real tasks instead of quizzes
You do the actual task in the actual interface. Get it wrong here, where it costs nothing, instead of on a live batch.
Get actual feedback
You don't just get a score. You get feedback on how you can improve.
The grade goes on your record
Each skill has a threshold. Clearing it goes on your record. Team visibility is planned.
Your credential follows you
It belongs to you and not to a company. Take it to the next project, the next human data company, the next lab.
Stay in touch with your network
Even the best projects and clients can sunset their work. Stay in touch with the people who made it memorable.
05 The proofThree numbers
We reduce reworks, churn, and onboarding time
Training platforms usually report completions and seat time. Neither predicts whether a batch gets accepted. These do, and they’re the same numbers whether you’re staffing the work or doing it.
Cohort rework rate · % of tasks
Time to qualified
Days from first lesson to first accepted production task. The number that decides whether you can staff a project next week or next month.
First-pass acceptance
Share of work accepted without revision. We measure it in practice first so you know the answer before the client does.
Rework rate
The margin killer. Tracked per person, per task type, per error class, and traced back to the lesson that failed to land.
06 Questions05 entries
The things people
ask first.
It does what an LMS does: courses, enrollment, progress, and analytics. But that's the delivery mechanism and not the point. The product is qualification: practice environments that mirror production workflows, scored against the rubric the work will be graded on. If all you need is compliance training, a general-purpose LMS is cheaper and you should use one.
We believe in upskilling annotators for free. Learn also lists more than 1,200 jobs, imported from the public careers pages of human data companies and marketplaces and refreshed regularly.
We work together with your team to compose a workflow.
Annotators own their record. Teams see it only when it is shared with them.
We first understand your needs. We scope out the work. We build out the rubrics and tasks. We run the course together. You get a list of annotators to hire.
Partnerwith us
You choose the research. We build the course, the sandbox, and the rubric, and we co-lead the session.