NerionX Multi-Tenant Commerce Hub (Medusa.js v2)

$5000-10000 USD

Awarded

Posted 2 days ago

$5000-10000 USD

Paid on delivery

We are seeking a technical partner to build NerionX, a platform designed on the principle: "Centralize Commerce, Decentralize Experience." By leveraging the modularity of [login to view URL] v2, we are architecting a "Hub" that manages global operational logic, while empowering autonomous "Spokes" (Brand Kits) to generate, localize, and store their own product content. We have a clear architectural roadmap and are looking for an expert team to help us execute a system that supports 50+ independent brands with zero shared-state bottlenecks. The project is structured into three distinct zones, ensuring a clean separation between global operations and brand-specific experiences: 1. The Hub (NerionX Core) The centralized engine and single source of truth for all operational and transactional data. Operational Core: A unified [login to view URL] v2 stack managing a master catalog of 25,000+ SKUs and variants. Global Governance: Enforcing "Price Floors" and unit-level profitability across all tenants via a centralized pricing engine and custom logic layers. Fulfillment Orchestration: A unified engine coordinating global inventory across multiple 3PLs with a strict 9-state lifecycle and pessimistic database locking to ensure inventory integrity. 2. The Spoke (Brand Kit) A containerized, deployable environment living on the Brand Owner’s infrastructure, responsible for the entire customer-facing layer. Local Commerce & Content: A lightweight Medusa v2 instance paired with Payload CMS 3.0. While the Hub provides raw product identities, the Spoke handles all localized titles, descriptions, and SEO metadata. Agentic Localization: A CrewAI (Python) or MCP connector for Claude Code / Codex pipeline that automatically generates brand-specific product content, which is then stored locally in the Spoke's CMS instance. Checkout Autonomy: Total independence to integrate regional payment providers (e.g., Stripe, Adyen) and manage localized pricing markups or subscription models. 3. The Integration Layer A hybrid communication model ensuring both transactional integrity and high-performance scaling: Asynchronous (Event Bus): Real-time broadcasting of raw product data and stock updates from the Hub to Spokes via Redis + BullMQ. Synchronous (Order Gateway): Direct REST API calls for finalized order payloads, ensuring the Hub validates fulfillment only after local payment success. The Technical Stack Summary Commerce Engine : [login to view URL] v2 (Node.js, TypeScript) Content Engine: Payload CMS 3.0 (Spoke-side only for localized content) Database: PostgreSQL 16+ (Centralized Core + Isolated Local Tenant Schemas) Search & Discovery: OpenSearch (Independent cluster for high-speed indexing) Frontend: [login to view URL] (App Router), Tailwind CSS, shadcn/ui Logic & Events: Redis, BullMQ, Python (CrewAI for Agentic Workflows)

Project ID: 40400958

About the project

115 proposals

Remote project

Active 7 mins ago

Looking to make some money?

Email address

Benefits of bidding on Freelancer

Set your budget and timeframe

Get paid for your work

Outline your proposal

It's free to sign up and bid on jobs

115 freelancers are bidding on average $7,272 USD for this job

@vladang

You’re essentially running a centralized commerce brain while each brand behaves like its own independent storefront—without letting inventory, pricing, or fulfillment drift out of sync. The real challenge is keeping that separation clean under load. I’d implement the Hub as a strictly authoritative service layer—catalog, pricing rules, and inventory guarded with pessimistic locking and idempotent operations to avoid race conditions across 3PL flows. Spokes stay fully isolated with schema-level separation and event-driven sync (Redis + BullMQ), ensuring they never block each other. For the order gateway, I’d enforce validation contracts + replay-safe endpoints so payment-confirmed orders can’t corrupt state. One improvement: introduce event versioning + replay queues early—this prevents silent data inconsistencies as you scale past 50 brands. I’ve built similar multi-tenant SaaS systems (Vue/Node/Postgres) handling isolated tenants with shared operational cores and high-volume async processing. Happy to map this into execution phases. Q1: How are SKU conflicts/versioning handled across Hub → Spoke updates? Q2: What’s the expected order throughput per minute at scale? Q3: Do Spokes require offline resilience or fully real-time dependency on Hub events?

$7,500 USD in 25 days

5.0

(17 reviews)

5.1

@techsty2014

Hi, I will help build NerionX as architected: Medusa v2 hub for the 25k+ SKU catalog and pricing floors, Medusa + Payload spokes consuming events over Redis/BullMQ, the synchronous order gateway after local payment, and the OpenSearch indexing layer. On fulfillment, pessimistic locking across the 9-state lifecycle is correct for integrity but tends to bottleneck at 50+ tenants and many concurrent orders. A reservation-with-TTL pattern (commit-time row locks, released on cancel or expiry) keeps integrity without serializing fulfillment. Questions: 1) Is the pessimistic-locking call locked in, or open to a reservation/saga pattern? 2) Spoke schemas as isolated DBs, or schemas inside the central Postgres? 3) CrewAI vs MCP for localization, decision still open? Looking forward to discussing further. Best regards, Faizan

$6,500 USD in 7 days