I’m a Supply Chain and Data Analyst with 2+ years of experience across operations, inventory, and financial risk analytics. My work spans two very different environments, a small-to-mid operations firm in Hyderabad where I rebuilt Cash and inventory workflows from scratch, and a US-based tech company where I work on the analytics function. In both, the job was the same: take messy operational data and turn it into decisions that save money or prevent a problem before it lands.

I hold an M.S. in Business Analytics & Global Supply Chain Management from University of New Haven (GPA 3.94, Beta Gamma Sigma Honoree), and my capstone work produced a production-grade predictive model that flagged $70M+ in credit write-off exposure. That’s not a classroom exercise, that’s what happens when you apply causal inference correctly to a real financial dataset.

Right now I’m based in the US, actively looking for analyst roles in supply chain, data, or business analytics. I don’t just build dashboards, I find the number that moves the decision.

Jul 2025 – Present | USA

• Built executive KPI dashboards in Power BI and Excel tracking order flow, on-time delivery, supplier performance, and supply continuity, cut decision latency by 30% and surfaced exception visibility across 4 critical operational metrics
• Led cross-functional analytics initiatives across 6+ stakeholder groups (Supply Chain, Operations, Compliance, Engineering), delivering prescriptive action plans that improved issue resolution cycle time by ~25%
• Ran structured root-cause analyses (5-Why, Pareto, fishbone) on procurement bottlenecks, standardized fixes reduced repeat escalations by 20%
• Built demand forecasting insights from CRM pipelines to surface supply risks 2–3 weeks ahead of potential shortages
• Ran scenario and sensitivity analyses on supplier delivery trends and demand adoption curves to support proactive supply plan adjustments

Jan 2022 – Feb 2023 | Hyderabad, India

• Redesigned inventory allocation across SKUs using SQL and Tally ERP, reduced stockout risk by 60% and directly improved order fulfillment rates
• Built anomaly detection models to flag supply discrepancies and logistics deviations, improved issue identification accuracy by 35%, cut mean time-to-corrective-action
• Engineered statistical demand forecasting models (time-series, seasonality decomposition, MRP-aligned), reduced excess inventory carrying costs by ~15%
• Automated supply reporting pipelines in SQL and Python, eliminating 100% of manual Excel reporting
• Built supplier performance scorecards and total-cost-of-supply analyses, identified sourcing inefficiencies, reduced procurement variances by ~10%
• Standardized procurement workflows (PR to delivery receipt) across 3 sites, cut cycle times by 18%

Python Propensity Score Matching Ridge Regression Random Forest Tableau Azure AutoML pandas scikit-learn

The credit team at Fleetcor was applying proactive treatments to high-risk accounts but had no evidence the interventions were actually working or just coinciding with accounts that would have recovered anyway. Selection bias made a straight comparison useless.

I applied Propensity Score Matching to build a valid control group, then used Ridge Regression to isolate the Average Treatment Effect (−$785/account). A Random Forest classifier (precision-recall optimized) scored which treated accounts would still churn. Azure AutoML validated the 89% model accuracy. A Tableau executive dashboard made the findings actionable without re-running code.

The key insight wasn’t just the $70M, it was that treated customers only dropped 5–7% in spend. Early intervention preserved the customer relationship. That’s what made the business case for scaling.

→ Finance-and-Credit

Excel Queuing Theory (Little’s Law) Demand Forecasting Centre of Gravity KPI Framework Design

CT Transit’s corridor had no quantitative basis for fleet sizing or hub placement. Schedules were static. Maintenance crews were positioned by convention, not data.

I built 3-scenario demand forecasts (±10–15% confidence intervals) so management had a range to stress-test, not a single number to trust blindly. Applied Little’s Law to model exactly how much wait time each additional bus removes — identified diminishing returns at 8 buses, which is the number that went into the recommendation. Centre of Gravity analysis on breakdown incident data repositioned maintenance hubs, cutting deadhead mileage by 12%.

→ Public_Transport_Supply_Chain_Operations

---### Retail Supply Chain Quality & Cost Analytics — IKEA Case Study Excel Pareto Analysis Nonconformance Rate Trending SPC / DMAIC Total Cost Modeling KPI Design

A global multi-country supply chain was tracking quality failures as company-wide averages, which made the numbers look fine at the top while hiding that 20% of suppliers were driving 75–80% of all defects. Managing the average meant ignoring the outliers doing the damage.

I ran Pareto segmentation to surface that concentration, then nonconformance rate trending per supplier and route to separate chronic failures from one-off incidents. A total landed cost model translated quality targets into P&L impact, so the conversation shifted from “we need a supplier program” to “here are the 12 suppliers we fix first, and here’s the dollar value.” Applied SPC and DMAIC framework throughout.

→ Retail_Supply_Chain_Work

Python spaCy (en_core_web_sm) scikit-learn Flask Jupyter

Operational teams waste time on the lowest-value part of their workflow: reading unstructured text, categorizing it, and assigning priority. I built a system that does this automatically.

The spaCy pipeline handles tokenization, lemmatization, NER, and stop-word filtering. A scikit-learn classifier (precision-recall optimized across all classes, not just accuracy) takes the normalized input and returns a task category and priority tier. The whole thing runs in a Flask web app, not a Jupyter notebook, an actual browser interface. The NLP layer and classifier are independently swappable, which is what makes it extensible rather than a one-off demo.

This architecture is identical to what drives enterprise service-desk triage, CRM ticket routing, and operations workflow classification, just at smaller scale.

→ Task-Management-System-ML

Python scikit-learn Collaborative Filtering Content-Based Filtering Cosine Similarity Flask

Built a hybrid recommendation system: collaborative filtering (user-item preference similarity) as the primary engine, content-based filtering (item attribute cosine similarity) as the cold-start fallback. Deployed on Flask.

The design decision that mattered was doing sparsity analysis during EDA first, that’s what revealed the user-item matrix was too sparse for pure collaborative filtering and made the hybrid approach the right call, not just a safe one. The domain is games, but the user-item matrix architecture transfers directly to product recommendation, demand-driven assortment, and supplier matching.

→ Game_Recommendation_System

M.S. Business Analytics — Global Supply Chain Management University of New Haven, West Haven, CT — GPA: 3.94 / 4.00 Beta Gamma Sigma International Business Honor Society Supply Chain Risk · Demand Forecasting · Operations Analytics · Predictive Modeling · ERP/MRP & Planning · Global Supply Chain Management Aug 2023 – May 2025

B.S. Mathematics, Computer Science & Electronics Osmania University, Hyderabad, India — CGPA: 8.18 / 10 Jun 2018 – Nov 2021

DataCamp: SQL, Python, R · Azure AI Essentials · LinkedIn Learning: Excel & Business Analytics · Google Digital Marketing · Google Analytics

Currently in USA, Open to relocation Actively seeking roles — full-time, contract, or hybrid