Meridian Group — CFO Intelligence Platform

Project Overview

• Built a complete end-to-end CFO Intelligence Platform for Meridian Group — a fictional 4-division Indian conglomerate — simulating a real corporate finance environment from raw data generation through to a 6-page executive Power BI dashboard

• Designed to solve a real business problem: finance teams manually reviewing thousands of transactions at month-end close, with anomalies staying hidden in the data for days before detection

• The platform automatically detects every financial irregularity using a 2-sigma statistical methodology consistent with forensic accounting and fraud detection practice

The Data Pipeline

• Generated 47,515 rows of realistic synthetic financial data using Python, Pandas, NumPy, and Faker across 4 business divisions and 2 full fiscal years (2023–2024)

• Built 6 structured tables in a star schema — dim_date (731 rows), dim_business_unit (4 divisions), dim_account (8 P&L lines), dim_scenario (Actual, Budget, Prior Year), fact_financials (46,784 rows), and fact_intercompany (731 rows)

• Incorporated seasonality multipliers, 8% year-on-year growth, daily transaction granularity, and independent Budget scenario generation with realistic plan-vs-actual variance

• Deliberately planted 5 financial anomalies into the dataset to simulate real-world close errors and test the detection system end-to-end

SQL Server Analytics

• Designed and loaded the MeridianCFO star schema database in SQL Server using SQLAlchemy and PyODBC

• Wrote 6 production-grade analytical queries using CTEs and window functions covering revenue variance analysis, full P&L budget vs actual, rolling 12-month EBITDA trends, intercompany reconciliation, 2-sigma statistical anomaly detection, and missing accrual identification

• Applied NULLIF division-by-zero protection, account-type-aware variance direction logic, and dynamic string concatenation for human-readable audit recommendations

Power BI Dashboard — 6 Pages

• Page 1 — Executive Summary: 5 headline KPI cards, 4 division revenue cards, RAG status table, monthly revenue trend line chart, and Total Anomalies Detected card displaying 7

• Page 2 — P&L Statement: Full income statement matrix with 8 account lines, conditional formatting (green for favorable, red for unfavorable), and 3 interactive slicers

• Page 3 — Variance Analysis: Waterfall bridge chart, EBITDA bridge bar chart, budget accuracy scatter plot, and top variances table

• Page 4 — Business Unit Deep Dive: 24-month multi-division revenue trend, RLS-restricted divisional P&L, and 12 division-level KPI cards

• Page 5 — Anomaly Intelligence: Automated risk detection panel with 4 anomaly cards, intercompany reconciliation table, and 5 risk KPI cards

• Page 6 — Data Governance: Pipeline architecture diagram, dataset inventory table, and anomaly detection methodology documentation

DAX Measures Library — 40+ Measures

• Revenue: Actual, Budget, Variance, Variance %, YoY Growth, MoM Growth, YTD, Rolling 3-Month

• Profitability: Gross Profit, Gross Margin %, EBITDA, EBITDA Budget, EBITDA Variance, EBITDA Margin %, Rolling 12-Month EBITDA

• Cost: Total COGS, COGS % of Revenue, Total OPEX, OPEX Ratio %

• Variance: PL Actual Amount, PL Budget Amount, PL Variance Amount, PL Variance %, Variance Type Flag, Budget Achievement Rate %

• Anomaly: Budget Breach Flag, Missing Accrual Flag, Total Anomalies Detected, IC Mismatch Count, Total Unreconciled Amount, Division RAG Status

• All measures validated in DAX Studio against Python-calculated benchmark figures

The 5 Anomalies Detected Automatically

• 🔴 West Division — Product Revenue 3.4× above normal — October 2023 — HIGH RISK

• 🔴 North Division — COGS breach, gross margin dropped sharply — June 2024 — HIGH RISK

• 🟡 South Division — G&A expense 67% above normal for 3 consecutive months — Q3 2024 — MEDIUM RISK

• 🟡 East Division — Depreciation accrual posted as zero — December 2023 — MEDIUM RISK

• ⚠️ 91 intercompany transactions unreconciled across 3 months — ₹31,000 exposure

Key Financial Outputs

• Total Revenue Actual: ₹8.06 Crore

• Revenue Budget: ₹8.26 Crore

• Revenue Variance: -2.4%

• EBITDA: ₹2.59 Crore

• EBITDA Margin: 32.1%

• Gross Margin: 46.4%

• Budget Achievement Rate: 97.6%

• Total Anomalies Detected: 7

Technology Stack

• Data Generation: Python, Pandas, NumPy, Faker, SQLAlchemy

• Database: SQL Server Express, PyODBC, MeridianCFO star schema

• Analytics: 6 SQL queries — CTEs, window functions, 2-sigma statistical methodology

• Visualisation: Power BI Desktop and Service, DAX, Power Query, Row-Level Security

• Validation: DAX Studio

• Version Control: Git, GitHub

• Project Management: Jira Scrum — 6 Epics, 34 Stories, 6 Sprints