BERT Model Fine-tuning for Skill Classification
Fine-tuned multilingual BERT models for large-scale skill classification on a dataset exceeding 200k job descriptions. Worked on training, evaluating, and refining natural language processing models to optimize entity and relation extraction for a knowledge graph. Ensured accurate labeling and model fidelity to support automated job-skill matching in production environments. • Used large-scale labeled text corpora to enhance classification accuracy. • Performed entity and relationship extraction from unstructured textual data. • Participated in evaluation, validation, and fine-tuning loops for model optimization. • Maintained labeling workflows for continual pipeline improvement.