About this role
Mid Level SQL Annotator
- Job Type Remote
- Qualification BA/BSc/HND
- Experience 5 years
- Location
- Job Field ICT / Computer
Job Summary
- We are in need of a detailed SQL DATA ANNOTATOR to join our team. In this role, you won’t just be labelling images you will be the critical link between our raw database pipelines and the machine learning models that power our analytics platform.
- You will query complex datasets, identify edge cases, and annotate structured/unstructured data to improve our AI’s natural language understanding (NLU).
Responsibilities
- Write advanced SQL queries (using JOINs, CTEs, WINDOW functions, and subqueries) to pull precise trainings slices from data warehouse.
- Use SQL string functions (SPLIT) to flatten nested JSON blobs, logs, or free text fields into tabular rows ready for labeling.
- Apply CASE statements and conditional logic to normalize dirty data before they reach the annotation queue.
- Manually review queried data and assign granular labels.
- Write dynamic SQL rules to automatically pre-label high confidence data so human annotators only review the ambiguous.
- Build automated SQL audit queries that can scan the annotated dataset daily to flag duplicates, conflicting labels, null values, or out of distribution samples before they reach the ML Team.
- Partner with ML Engineers to test proposed annotation guidelines against actual database records.
Requirements
- Bachelor’s Degree in Computer Science or a relating field.
- 5+ years’ experience writing advanced SQL in a professional setting and can be able to identify the difference between INNER JOIN, LEFT JOIN, and LATERAL FLATTEN without Google.
- Experience with annotation tools (Label box, Super Annotate) and familiar with python or R for Data manipulations.
- Ability to spot a missing comma or a mislabeled data point from a mile away and understand the ripple effect it has on model performance.
- Ability to translate technical database structures into plain English annotation guidelines for non-technical team members.