Financial Simulation & Customer Segmentation for a Grocery App

The Challenge

The executive board, Sales, and Finance teams needed a single reliable view of the business, but payout calculations for agents and drivers were done ad hoc, and marketing had no way to target customers beyond broad segments.

The Solution

• Owned the end-to-end BI system on Metabase and Superset, running ETL on the data warehouse for real-time reporting to the board, Sales, and Finance.
• Built financial simulation tools in Python to compute agent wages, driver salaries, and promotional budgets against company targets.
• Applied K-Means clustering on demographic and behavioral data to segment customers for sharper marketing targeting.
• Designed fraud-prevention classification rules and tooling to flag suspicious activity.

The Result

• Payout calculations became consistent and auditable, improving fairness for agents and drivers while keeping spend within budget.
• Marketing gained data-backed customer segments instead of one-size-fits-all campaigns.
• Reduced exposure to fraudulent activity through proactive flagging rules.

Stack