Development of MarginGuard Platform

MarginGuard is a SaaS platform for finance and operations teams to understand profitability, identify margin leakage, investigate its causes, and track corrective actions.

I built the platform with Next.js and FastAPI, connecting data ingestion, profitability calculations, leak detection, AI-assisted investigation, and action tracking.

Growing revenue does not always mean growing profit. Supplier cost increases, excessive discounts, refunds, and operational expenses can gradually erode margins.

When business data is spread across spreadsheets and separate systems, teams can struggle to identify where profitability is deteriorating, understand why, and decide what to address first.

MarginGuard was built to connect those steps and keep the evidence behind each finding accessible.

Data ingestion: Upload CSV files, map fields, preview records, and review validation errors and rejected rows.

Profitability analysis: Explore revenue, costs, and margins across products, customers, and sales channels.

Margin leak detection: Identify deteriorating margins against configured baselines and thresholds, with estimated financial impact and supporting evidence.

AI-assisted investigation: Generate explanations and recommendations, and ask follow-up questions through an investigation assistant.

Action and recovery tracking: Convert recommendations into assigned action items, track progress, and evaluate margin recovery against a baseline.

Team workspaces: Support multiple organizations, role-based permissions, shared settings, and audit history.

The core challenge was keeping financial calculations consistent from imported records through dashboard summaries, detected issues, and recovery evaluations. Data validation, normalization, and baseline comparisons needed to follow clear rules so findings remained traceable to their underlying evidence.

The platform also required strict separation between organizations. Organization-scoped data access and role-based permissions control which records and actions each team member can access.

Another challenge was integrating AI into a financial workflow. Profitability calculations and detection rules run in application code, while AI explanations and investigations use structured evidence from the workspace. Unavailable states allow the core workflow to remain usable when an AI response cannot be produced or validated.

A working platform that connects business data with profitability analysis and tracked corrective actions. Users can move from a margin overview to a specific finding, inspect its supporting evidence, and assign a response within the same workspace.

Built as an independent portfolio product using sample business data, MarginGuard demonstrates full-stack SaaS development across multi-tenant architecture, background processing, financial analytics, AI integration, and operational workflows.

My role: Full-Stack Developer Technology: Next.js · React · TypeScript · Tailwind CSS · Python · FastAPI · PostgreSQL · SQLAlchemy · Redis · Gemini