Samiksha Sharma

Collaborated with a faculty supervisor on applied data science research, focusing on AI-driven modelling and predictive analytics.

• Applied YOLOv5 and Vision Transformers to high-resolution drone imagery for crop yield estimation, achieving mAP of 0.82, supporting data-driven agricultural planning.
• Designed end-to-end computer vision pipelines, including data preprocessing, model training, evaluation, and inference on geospatial imagery.
• Developed visual analytics dashboards and geospatial visualizations to translate model outputs into interpretable yield insights for non-technical stakeholders.
• Collaborated with academic researchers to align model design with research objectives, ensuring methodological rigor and reproducibility.

Worked with global retail and consumer goods clients to design and implement machine learning and analytics solutions that improved customer experience, pricing strategy, and supply chain efficiency.

• Built predictive models in Python (scikit-learn, PySpark) to forecast customer churn and purchasing trends, improving targeting precision by 15%.
• Developed machine learning models to analyze price elasticity and promotion performance, helping client teams increase promotional ROI by 10%.
• Designed automated time-series forecasting pipelines that integrated POS and inventory data, reducing stockouts and excess inventory by 18%.
• Migrated legacy retail systems to AWS and Azure, improving data refresh efficiency by 40% and enabling near real-time analytics.
• Delivered interactive dashboards (Power BI, Tableau) to monitor sales and inventory KPIs across 50+ regions, reducing manual reporting time by 30%.
• Partnered with marketing, operations, and data engineering teams to translate analytical insights into actionable retail strategies.