Inventory write-off 35% ↓ — $1 million impact
Created forecasting model for 2K+ appliance spare parts leveraging policy claim rates, part failure rates, SKU categorization by volume & demand variability pattern.
NPS 6 pts ↑
Conducted sentiment analysis and topic modeling on web-scraped social media data from 5 platforms to identify key customer issues and recurring pain areas.
Last-mile cost $600K ↓
Classified addresses into high, medium, and low categories based on text attributes, blocking low quality entries daily during order placement and aiding last mile delivery
Last-mile cost $9 million ↓
Optimized delivery efficiency by identifying low-intent customers and implementing strategy to halt attempts after a threshold and prioritizing high-success orders.
Pickup failure losses $1 million ↓ — Speed 6% ↑ — RES 4 pts ↑
Used process mining to identify and address operational bottlenecks to reduce costs and improve customer experience.
Created blueprints for inbound, storage, and dispatch layouts, rack placement for 100,000 sqft warehouses and express DCs using models based on volume flows, sales trends, and demand forecasts.
Packaging density 8% ↑ — Packaging cost $120K ↓
Developed algorithm to determine optimal box/bag size for customer orders based on contents, implemented real-time via API for 10 million annual orders
Margin 0.2% ↑ — $500K impact
Created an algorithm to determine delivery charges based on source - destination, courier rates, and business constraints, optimizing cost while preserving SLA and conversion.
Picking TAT & walking distance 13% ↓
Engineered a methodology to optimize product placement within warehouses based on sales trends, expiration dates, and warehouse-specific constraints
Expired inventory cost $350K ↓ — Picking TAT 7% ↓
Automated real-time inventory monitoring, bin replenishment task creation and audits, ensuring inventory sanity and physical-digital sync.
$50 million sales ↑
Segmented world’s largest e-comm customer base of 250 million+ based on transactional and demographics data, leveraging survey responses using a Random Forest model.
$5 million cost ↓
Built a machine learning model for a Swiss multinational to predict order cancellations using weather forecasts, streamlining labour planning and raw materials
Split factor 28% ↓ — $15 million impact
Network design for America’s largest Arts & Crafts company, using demand clustering, network optimization, SKU categorization on COGS and MBA, inventory profiling.
Developed Normal, Synthetic and Sequential Affinity models on R to identify product combinations that have higher likelihood to be purchased together to promote cross-selling online and plan in-store rack product placements
For the purpose of reviewing monthly performance of the world’s largest retailer with monthly sales upwards of $25 billion, formulated back-end on Google Cloud Platform (GCP) and front-end on Metabase of a series of dashboards; which are used by store managers and executive management of the company alike to make decisions cumulatively impacting billions of dollars.
Analysed the behavior of customers of retailer A in competitor retailer B and vice versa for 40 retailers by comparing KPIs like Spend per Customer, Orders per Customer and Retention at a weekly/monthly, cumulative and yearly aggregated level.
Built hourly updated dashboard showing latest trends across 21 dimensions (department, demographics, device type, shipment type etc) during Black Friday - Cyber Monday for quick strategy building.