Skip to main content

Process

From model objective to dataset delivery

The process starts with what the data needs to teach, represent or measure.

01

Define the problem

Tell us what the model needs to do, which modality matters and what conditions the data needs to represent.

02

Design the dataset

We define the useful record, the source media, the supervision, the metadata and any evaluation structure the task depends on.

03

Build the collection

The required data is produced or prepared according to that structure, with the relationships between files, records and modalities preserved where they matter.

04

Package the data

The final dataset is organised into the agreed delivery format, connecting the source media to the records and technical metadata your team needs to work with it.