Data · Risk · Analytics · Operations
I build analytical solutions that turn complex transactional and operational data into structured insights, risk signals, and decision-support tools.
My work spans customer analytics, risk and fraud analytics, statistical analysis, data quality, and operational monitoring. My projects combine analytical methods with practical implementation, from customer and risk analysis to statistical validation and data-quality systems.
- Customer & Promotional Analytics
- Risk & Fraud Analytics
- Statistical & Exploratory Analysis
- SQL & Relational Data Analysis
- Data Quality & Validation
- Decision Support & Dashboarding
- Data-Driven Operations
An end-to-end customer analytics solution using the dunnhumby Complete Journey dataset, analyzing 2.6M transactions across 2,500 households representing $8.06M in sales to identify targeted promotional opportunities.
Key Highlights
- Household-level promotional opportunity scoring
- Customer segmentation and category affinity analysis
- Methodology validation and scoring refinement
- Identification of zero-inflation and department-size biases
- Analytical workflow from Python to PostgreSQL
- Cross-validation using Python, PostgreSQL, and AWS Athena
- Three-page Power BI dashboard and recommendation memo
- Identified 10 highly engaged households with strong category affinity and zero recorded campaign history
- Quantified an illustrative $18.1K sensitivity case, explicitly separated from evidence-based findings
Tech Stack
Python SQL PostgreSQL AWS S3 AWS Athena Power BI
Repository
➡️ https://github.com/Saksham3124/dunnhumby-customer-promotional-opportunity
An end-to-end invoice analytics and vendor risk solution analyzing 50K+ GST invoices across 210 vendors to detect anomalies, quantify vendor risk, and support audit prioritization.
Key Highlights
- Four-layer risk analytics framework
- Rule-based validation and statistical anomaly detection
- Z-Score & IQR-based analysis
- Weighted vendor risk scoring using four risk signals
- 7,500+ suspicious invoices identified (15%)
- 32 vendors classified as HIGH-risk
- Gemini API-based evidence-grounded risk narratives
- 32 vendor narratives connected to an interactive Tableau dashboard
- Evidence-grounded AI explanations that preserve deterministic risk scores and tiers
Tech Stack
Python SQL PostgreSQL Tableau Gemini API
Repository
➡️ https://github.com/Saksham3124/gst-invoice-anomaly-detection
An analytical data-quality and reliability system processing 10,089 E-Way Bill records across FY2022–23 and FY2023–24, combining validation, reconciliation, and statistical analysis.
Key Highlights
- Data-quality validation and cross-table reconciliation
- KS and PSI-based year-over-year statistical analysis
- 204 statistical results across multiple dimensions
- Automated reliability checks and PASS/FAIL validation
- 7 controlled corruption scenarios
- 125 automated tests covering missing records, duplicates, invalid values, structural defects, and a controlled INR 500 Cr alteration
- Airflow-orchestrated validation and PostgreSQL-based trusted storage
Tech Stack
Python SQL PostgreSQL Apache Airflow Docker SciPy Streamlit
Repository
➡️ https://github.com/Saksham3124/ewaybill-data-reliability-pipeline
| Category | Technologies |
|---|---|
| Programming | Python, SQL, MATLAB |
| Data Analytics | Pandas, NumPy, SciPy, Statistical Analysis, EDA |
| Databases | PostgreSQL |
| Analytics & BI | Power BI, Tableau, Streamlit, Excel |
| Cloud | AWS S3, Athena |
| Data & Engineering | ETL/ELT, Data Pipelines, Data Quality & Validation, Apache Airflow, Docker |
| Tools | Git, Jupyter, APScheduler |
- Product and customer analytics
- Risk and statistical analysis
- Data quality and analytical reliability
- Cloud-based analytics and data workflows
- Operational and decision-support systems
- LinkedIn: https://linkedin.com/in/kumarsaksham/
- Email: kumarsaksham560@gmail.com