I map howcompanies actuallywork - then getpeople using AI.

The problem

The full franchise lifecycle - warranty registration, claims, grievances, onboarding, product catalog, announcements, POSM distribution - was running across disconnected manual processes, which made it slow to operate and easy to abuse.

What I built

• Built and deployed a production-grade platform covering the entire franchise lifecycle under one role-gated system for Customers, Franchise Partners, and Admins.
• Designed a multi-layered fraud detection system combining GPS distance verification against a store database with camera EXIF and metadata analysis to flag fraudulent warranty submissions.
• Consolidated warranty, grievance, onboarding, catalog, announcement, and POSM workflows into a single operational surface with full admin visibility.

The problem

Sales, Leads, Finance, Export, Production, and OE Dealer Network data each lived in its own spreadsheet, so every cross-team question required manual reconciliation before anyone could answer it.

What I built

• Architected and shipped a full-stack MIS consolidating six departments into a single role-gated dashboard, replacing scattered spreadsheets across the organization.
• Built automated data pipelines syncing live Google Sheets and Excel uploads into a structured Postgres backend with FastAPI and SQLAlchemy, eliminating manual cross-team reconciliation.
• Building an integrated AI chatbot on Claude and GPT-4o with direct database access, applying agentic system principles so responses are grounded in live company data rather than a static knowledge base.

The problem

Nobody could describe the dealer order process end to end. It crossed four functional teams, and the handoffs between them were where orders stalled and data drifted.

What I built

• Ran an end-to-end workflow analysis of the dealer order process across four functional teams - ASM, Production, Accounts, and Dispatch - documenting all 44 process steps.
• Identified the automation gaps and data-integrity risks hiding in the handoffs between teams, and used that map as the specification rather than starting from assumptions.
• Now designing and building the system from those findings, covering ASM approval, production tracking, invoicing, and dispatch.

The problem

Customers had no fast way to confirm whether a store selling Autoform products was an authorized franchise, which left the aftermarket brand exposed to counterfeits.

What I built

• Built an anti-counterfeit web platform letting customers scan a QR code to verify authorized Autoform India franchise locations in real time.
• Automated batch QR code generation and CSV-to-JSON data pipelines with custom Node.js scripts, eliminating manual data entry for 400+ franchise stores.