Retail Analytics and ERP Intelligence Platform
- Role
- Solo developer
- Built with
- Type
- Odoo ERP analytics with AI

Screens
About the project
A full-stack retail analytics platform that turns a live Odoo ERP into a real-time decision system, with FIFO inventory valuation, sales and workforce KPIs, and AI-generated insights grounded on real data.
Role: Solo developer, about 274 of 280 commits, 10 months and ongoing.
Designed and built a full-stack retail analytics platform that turns a live Odoo ERP into a real-time decision system. It mirrors ERP and e-commerce data into a single source of truth, then layers FIFO inventory valuation, sales and finance KPIs, workforce analytics and AI-generated insights on top. One back end and one SPA serve operations, finance and HR.
What I delivered
- AI insight engine: retrieves and aggregates the relevant operational data (attendance, overtime, sales) from Postgres, augments an LLM prompt with that grounded context, and returns schema-constrained JSON: trend prediction with confidence, root-cause hypotheses, chronic-risk scoring and executive summaries. Grounding on real data keeps outputs auditable and hallucination-resistant.
- Document AI: Azure Document Intelligence (prebuilt invoice model) extracts structured line items from uploaded PDF and image invoices, feeding the procurement pipeline.
- ERP sync layer: read-only Odoo integration over XML-RPC with an entity-first, ordered cron pipeline (products, categories, suppliers, orders, inventory, accounting) and per-entity sync-state tracking.
- E-commerce channel: Salla integration with OAuth (Passport), REST product and order sync, and webhook event handling as a downstream sales channel.
- FIFO inventory valuation engine with stock valuation layers, quants, and daily and intraday snapshot facts for point-in-time stock and margin analysis.
- KPI and target system: sales and finance KPIs, per-team and per-category targets with full audit history and month-to-date pacing.
- Workforce analytics: attendance engine with schedules, shifts, weekly patterns, a violation-detection engine and leave management.
- RBAC: role and permission model enforced on both API and UI.
- Ops tooling: cron management dashboard, WhatsApp notifications, spreadsheet import and export (XLSX, CSV), and a Swagger-documented REST API.
Scale: 20 controllers, 20 route modules, 56 data models, Odoo, Salla and Azure integrations, Dockerised with CI/CD auto-deploy.
Stack
- Node.js
- Express
- Prisma
- PostgreSQL
- React (Vite)
- JWT + Passport
- Odoo XML-RPC
- Salla REST/OAuth
- LLM structured output (RAG-style grounding)
- Azure Document Intelligence
- node-cron
- Zod
- Swagger
- Docker

























