Data quality lead

Pay in the destination countries

Monthly, in US dollars. Each country is an Imora admission package — open one for that route's fees, language requirement and deadline.

What does the day look like?

Your day starts inside queues in Labelbox, CVAT, Prodigy, Label Studio, Scale Studio, SamaHub, SamaIQ and Kili Technology, auditing batches across computer vision, natural language processing and Reinforcement Learning from Human Feedback projects. You calculate Inter-Annotator Agreement with Cohen's Kappa or Fleiss' Kappa, precision, recall, F1-score, Defect Rate and First Time Right, place hidden golden questions in queues, review contested audit grades to set final ground truth, and remove annotators below the 95% to 99.5% accuracy threshold.

40% of the day is direct inspection of labeled datasets, edge-case checks and golden set construction. 60% is SOP and grading rubric writing, tracking in Jira, Linear, Asana and Notion, throughput measured as Units Per Hour in SQL, Google BigQuery, Microsoft Excel and Google Sheets, messages in Slack, Discord and Loom, and calibration calls every 2 weeks with client Machine Learning engineers. You work 40 to 50 hours per week on a 24/7 async operation across Kenya, India, Philippines and Latin America, with late-evening or early-morning calls with US West Coast or European teams and async channels monitored outside local business hours.

Where does this lead?