AIML engineer
Audio
Text GenerationI worked on AI training data preparation and quality evaluation tasks focused on improving large language model performance. The project involved reviewing, analyzing, and validating multilingual and speech-related datasets, including translated text, text-to-speech (TTS) outputs, conversational responses, and structured annotation workflows. I also worked on identifying inconsistencies, improving data formatting, checking alignment between inputs and outputs, and helping optimize data pipelines used for model training and evaluation. The project handled large-scale datasets across multiple tasks, where maintaining consistency and annotation quality was critical. Quality measures included following strict annotation guidelines, validating outputs through repeated reviews, checking edge cases, ensuring accurate labeling, and maintaining high attention to detail during evaluations. I also collaborated on debugging data-related issues, improving workflow efficiency, and documenting findings clearly for internal reporting and model improvement discussions.