Maansi Bharti

I led the Generative AI module for a plant disease detection and treatment project, building the data pipeline for a CNN across 15+ categories and 5,000+ labeled images. The model achieved 92% validation accuracy, while compression and quantization improved inference speed by 25%.

I've built Python-based processing scripts that turn model outputs into structured, actionable recommendations. I enjoy translating business problems into data-driven solutions, validating data quality, and collaborating with stakeholders on reliable deployment.

In a customer churn prediction project, I cleaned and transformed 10,000+ records, developed classification models with 87% accuracy, and improved robustness by 15% through hyperparameter tuning. I also built real-time hand-tracking pipelines sustaining 30+ FPS and applied statistical analysis and logistic regression to Titanic survival prediction.

I'm a B.Tech Information Technology graduate with experience in Python, SQL, ETL/ELT workflows, data modeling, machine learning, and relational databases. I'm eager to apply this foundation to modern data platforms including Snowflake and Databricks.