Jr Data Scientist
Current• Enhanced vector embedding accuracy by 40% by fine-tuning the GTE Large 1.5 model, leveraging CoSENTLossto optimize ranking of JD and CV pairs based on semantic similarity for increased matching performance.• Reduced processing time by 75% by bench marking embedding models on accuracy and performance metrics,using average vector pooling during training and evaluating results with Spearman correlation.• Created chat bot functionality to display job listings based on user prompts and location, increasing websitetraffic by 30%. Built an API to call an LLM and generate queries, retrieving relevant jobs from a SOLR database.