I turn messy data into decisions that move the business
Product Data Scientist with 6+ years in ML and 9+ years in analytics across telecom, retail, and fintech
I build systems that help businesses decide what to sell, who to target, and when to act.
Built systems that improve retention, optimize inventory, and drive revenue
What I do
Forecasting
Predict demand and customer behavior
Decision Systems
Translate predictions into business actions
Analytics
Turn raw data into clear insights
Experience
Recent roles
Nul Global - Senior Data Scientist
Built end-to-end demand forecasting systems, reducing retail overstock by up to 80%.
ATOM Telecom - Senior Manager (Advanced Analytics)
Built churn prediction and recommendation systems, generating $0.2M revenue and saving $0.4M through retention optimization.
Ghostcap - Analytics Manager
Improved onboarding funnel (+20%) and automated BI reporting, reducing manual work by 65%.
Workflow
From messy signals to confident inventory calls
The system ties together sales telemetry, promo plans, and supply constraints into an always-on workflow that feels like a product, not a reporting deck.
System diagram
Productized capabilities
Opinionated modules expose the system like a product surface so teams can plug it into real planning cadences.
Signal ingestion
Sales, product, and promo feeds normalized with governance baked in.
Scenario forecasting
Forward-looking demand simulation with knobs for promos or supply shifts.
Decision delivery
APIs and playbooks that push guidance into inventory + replenishment tooling.
System Detail
Inside the build
A compact stack that moves clean inputs through modeling to decision outputs.
Data foundation
Clean sales, product, and promo layers modeled for downstream demand work.
Demand forecasting
Temporal models plus scenario logic that stay aligned with business cadence.
Decision outputs
Structured guidance, alerts, and playbooks that plug into planning tools.
Business Impact
Why this system matters
Every release is measured against inventory efficiency, service levels, and the speed of operator decisions.
Impact
↓ Overstock
Less capital trapped in slow movers.
Impact
↑ Availability
Shelves stay ready for demand spikes.
Impact
↑ Decision speed
Operators act faster with pre-built guidance.
Case Study Preview
Forecasting and Replenishment Decision System
A condensed look at how the workflow is applied end-to-end for a merchandising org.
Problem
Promo-heavy categories were oscillating between excess stock and missed sales because planning relied on backward looking spreadsheets.
Approach
Built a signal ingestion layer, scenario-aware forecasts, and a decision surface that routes replenishment guidance into weekly buyer rituals.
Contact
Let's connect
Quick ways to reach me about product data science work.