Outlier
Large-scale AI training project supporting model development across multimodal datasets, including text, image, and audio. Work involved structured data labeling tasks such as bounding box annotation, object detection, named entity recognition (NER) classification, question answering, emotion recognition, text summarization, transcription, and audio recording, along with evaluation and rating of model outputs and prompt-response writing for supervised fine-tuning (SFT). Contributions also included text generation and consistency checks to improve training data quality and model performance. Tasks were completed within defined production workflows and labeling guidelines, with adherence to strict quality measures, including protocol compliance, accuracy verification, and internal review standards across a high-volume project environment.