Skip to content

Expert data and evaluation programmes

Specialist contribution is shaped around the quality of the signal and its usefulness to model development.

Discuss the programme family
Specialist engineering programme
Provenance + source corpusOrigin retained

Select the right programme within expert data and evaluation programmes.

Capability dossier

The programme family

Specialist contribution is shaped around the quality of the signal and its usefulness to model development. Softinator connects specialist contribution to the engineering, information and operating requirements around the task.

Choose a focused engagement

The catalogue includes advanced mathematics training data, scientific training data, expert code training data, multimodal instruction data, preference and critique data. Each programme has its own contribution, artefacts and acceptance scope. Use the programme links below to identify the relevant engagement.

Commission the work

Bring the operating task, selected environment and result that matters. The engagement establishes scope, access, delivery responsibilities and the evidence used to assess the outcome. Connected platform and project pages provide the relevant context.

Choose the programme.

Advanced mathematics training dataCommission specialist-authored problems, solutions and review for model development.Agent trajectory dataCapture meaningful sequences of decisions, tool calls and task outcomes.Computer use task dataCreate application interaction tasks with executable goals and reviewable outcomes.Data quality and deduplicationImprove training-corpus integrity through duplicate, coverage and quality controls.Dataset provenance engineeringMaintain rights, versions and lineage across training and evaluation inputs.Domain expert evaluationAssess model behaviour using specialist judgement and an agreed evaluation rubric.Evaluation contamination auditsCheck overlap and leakage risks between training data and evaluation material.Expert code training dataCreate engineering tasks and reviewed code artefacts for code-model development.Multilingual corpus engineeringCurate multilingual corpora with provenance, quality and domain coverage.Multimodal instruction dataConnect images, language and task outcomes in a structured model-training corpus.Preference and critique dataCollect discriminative judgements and critiques tailored to model failure modes.Rubric and verifier designDefine task-success criteria and reviewable verifiers for complex model outputs.Scientific training dataBuild domain-specific contributions across scientific reasoning and interpretation tasks.Synthetic data validationValidate generated examples before they enter domain training programmes.

Follow the connection.

Explore the method, its context
and the engineering underneath.

Trust centre

Enterprise procurement

Commercial scope, technical requirements and the evidence needed to commission the work.

Bring the next
requirement.

Discuss your programme