Omdena Gun violence
Text
ClassificationOver my time working in data labeling and AI training, I have focused heavily on content moderation, safety compliance, and data filtering projects. My most notable project involved large-scale text annotation for a prominent children's website model. The primary objective was to build a robust safety filter capable of detecting and blocking profane, explicit, or age-inappropriate language. This required analyzing complex text strings, identifying masked or deliberate misspellings of profane words, and evaluating the context of conversations to flag subtle cyberbullying or safety risks without over-filtering benign, natural speech. Through this project, I gained deep expertise in handling nuanced datasets where context is everything. I am highly proficient in applying strict taxonomies and multi-layered annotation guidelines to ensure excellent inter-annotator agreement. My experience bridges the gap between raw user data and high-quality, human-labeled inputs, directly helping engineering teams establish tight safety guardrails and build highly reliable, ethically aligned AI models.