Data Scientist Intern
CurrentApplied unsupervised ML techniques (probabilistic models, clustering) to model search success and user behavior for the company’s largest product, improving Search and Product Analytics metrics.Experimented with embedding models (GPT, Gemini, LLama), unsupervised approaches, and A/B testing, boosting the model performance by 20% and identifying optimal strategies for ML models.Leveraged Large Language Models (LLMs) to validate models and improve overall model performance.Designed and implemented an Anomaly Detection Pipeline to identify data drifts and out-of-distribution shifts.Migrated models and pipelines into Production on Azure, ensuring reliable and scalable deployment.Refined metric design by leveraging statistical and mathematical approaches, improving existing KPIs, and creating new metrics for tracking search experience and customer satisfaction.
