Mahavia U

Data Science - Analyst

LinkedIn • Período integral

Aug 2025 - Aug 2026 • 1 yr

Built and evaluated predictive models on large real world datasets, working end to end from raw data through cleaning and feature preparation to training, testing and performance evaluation. Worked primarily in R with Tidymodels and XGBoost, and assessed how model performance held up across different segments rather than relying on a single headline accuracy figure. Delivered segment level performance analysis across categories and regions, comparing groups on the metrics that mattered and identifying where the differences were real. On one project sense checking revealed that older records were distorting the comparisons, and correcting for it reversed the conclusion the business had been working to. Careful validation before reporting is a standard part of how I work rather than an extra. Produced customer segmentation from large administrative datasets to support a targeted campaign, grouping customers by behaviour and status to show where activity would be most effective. Delivered analysis end to end under tight deadlines, then presented the findings and written reports directly to senior stakeholders and specialist teams in language they could act on. Tools: R (Tidymodels, XGBoost, ggplot2, R Markdown), SQL, SAS, Git, Excel.

Financial Data Analyst (Internship)

Bupa UK • Período integral

Jun 2022 - Aug 2022 • 2 mos

Led a cross functional team of four interns from Marketing, Finance and Clinical over a six week project investigating consultant fee queries. Benchmarked Bupa consultant fees against competitors on a like for like basis, which showed clinician fees were being set below the market benchmark and prompted a senior management fee review.