Ranking + Recommendations at Scale (Walmart)

Role

Lead Data Scientist (Strategy)

Timeline

12 Months

Team

Strategy Directors, Network Engineers, Marketing Leads

My Focus

Geospatial Analytics, Causal Inference, Strategic Planning

PySparkGCPBERTRankingRecommendation SystemsSQL

Business Impact

+10% CTR; +25% Recall@K

Scale

Billions of transactions

The Challenge

The Challenge: The Infrastructure Risk

Deploying 5G infrastructure is capital intensive. Huawei needed to ensure that every new tower location would generate maximum ROI across Latin America.

The Bottleneck: Traditional planning relied on simple population density data, which failed to account for purchasing power or "willingness to pay."
The Goal: Develop a precision-targeting engine to identify high-value "Lifestyle Zones" rather than just high-population areas.

The Architecture

I built an end-to-end Market Mix Model that combined geospatial data with economic behavior:

Data Ingestion: Integrated OpenStreetMap data with internal network usage logs to create a rich geospatial dataset.
Lifestyle Clustering: Developed a custom "Lifestyle Zones" algorithm using K-Means and DBSCAN to segment areas based on commercial activity and housing types.
Optimization: Applied Linear Programming and XGBoost to predict demand and recommend optimal site placements under budget constraints.

System Architecture Diagram

graph TD
    A[Geographic Data<br/>OpenStreetMap] --> B[Clustering Algorithm<br/>K-Means + DBSCAN]
    B --> C[Lifestyle Zones<br/>Segmentation]
    C --> D[XGBoost<br/>Demand Prediction]
    D --> E[Linear Programming<br/>Site Optimization]
    E --> F[Pricing Model<br/>Elasticity-Based]
    F --> G[PowerBI<br/>C-Level Dashboard]

    H[Demographic<br/>Data] --> C
    I[Network<br/>Coverage Data] -.->|Constraints| E

    style A fill:#0066ff,stroke:#0052cc,stroke-width:2px,color:#fff
    style B fill:#4C9AFF,stroke:#0066ff,stroke-width:2px,color:#fff
    style C fill:#0066ff,stroke:#0052cc,stroke-width:2px,color:#fff
    style D fill:#4C9AFF,stroke:#0066ff,stroke-width:2px,color:#fff
    style E fill:#0066ff,stroke:#0052cc,stroke-width:2px,color:#fff
    style F fill:#4C9AFF,stroke:#0066ff,stroke-width:2px,color:#fff
    style G fill:#0066ff,stroke:#0052cc,stroke-width:2px,color:#fff
    style H fill:#666,stroke:#444,stroke-width:1px,color:#fff
    style I fill:#666,stroke:#444,stroke-width:1px,color:#fff

The Impact

The insights from this model directly influenced the deployment strategy for the 5G rollout.

Metric	Legacy Planning	AI-Driven Strategy
Targeting Method	Census/Population Density	Behavioral "Lifestyle Zones"
Precision	City/Regional Level	Hyper-local Block Level
Efficiency	Standard Conversion	+39% Targeting Efficiency
Financial Value	Baseline Revenue	+$40M Incremental Revenue

Collaboration & Strategy

This project was not just about code; it was about influencing executive strategy:

Stakeholder Management: I presented findings directly to C-Level executives via interactive PowerBI dashboards, translating complex clustering logic into actionable "Go/No-Go" maps.
Cross-Market Adoption: The "Lifestyle Zones" methodology was so successful it was adopted as the standard planning framework across multiple Latin American markets.

Interested in similar solutions?

Let's discuss how we can build scalable ML systems for your business challenges.

View More Case Studies