Approach To Quality

Quality At Source

Understanding your data quality framework, data governance philosophy and model design is important to us.

Starting with your data quality framework forces us to look beyond your current needs, view your requirements from different perspectives and puts us in the mindset to creatively address gaps in your model performance. 

Setting the Standard: Measuring data quality starts with setting the standard for the best evaluators, to ensure they meet use-case qualification criteria before being assigned to deliver your local relevance, machine intelligence or model optimization solution.

Our expertise at connecting with and qualifying the best evaluator talent around the world is the starting point of your data quality journey with Peroptyx. 

Sourcing and Scaling: Quality starts with setting the evaluator acceptance threshold at the 90th percentile of all applicants for our data evaluator positions. This is the entry point into the first stage screening process for all evaluation teams. We always balance this threshold with baseline qualitative and quantitative expectations for successful applicants.

Use-Case Qualification: We test each successful candidate for their theoretical and practical competence relative to the use-case or solution being designed and deployed. The qualification process can range from customer-determined criteria to our own independent criteria designed in consultation with your team.

Generally speaking, the overall minimum qualification score is in the 85%-90% range.

Blind Evaluation: We continually assess the standard of work by running both regular and random 'blind evaluations', using a predefined gold standard set of results to set the perfect benchmark score. 

Our best evaluators consistently score above 90% on the most difficult assignments. 

Moderator Review: Our Moderators review results from blind evaluations and apply a causal-based selection methodology to drive data quality and model improvements. This process transforms evaluator feedback and insights into actionable outcomes.

Data Science QC Analysis: Our Data Science and Quality Control experts deep dive into the operational and quality data to uncover performance insights, improve overall solution value and work as a scalable extension to your management, data science and engineering teams.

Be Part of the Peroptyx Community

Evaluator Application