Live Inventory Aggregator & AI Quote Engine

We need an expert to build a live inventory aggregator and AI quote engine MVP for RehabEquip AU. The platform will help Australian occupational rehabilitation consultants source ergonomic equipment for injured workers. The core workflow requires a restriction wizard with structured selection fields that map to Australian clinical metrics, including material handling limits, postural tolerances, and upper limb constraints. The solution must support one manual entry interface, integrate with live inventory data, and generate accurate quotes based on selected restrictions. Please review the full project brief and business rules before applying. 1. Project Core Logic & The User Workflow The platform streamlines how Australian occupational rehabilitation consultants source ergonomic equipment for injured workers. It requires only one manual entry interface: an FCE (Functional Capacity Evaluation) Restriction Wizard. You will build this using structured selection fields (dropdowns, numeric inputs, and multi-select checkboxes) that map strictly to Australian clinical metrics: • Material Handling: Maximum occasional and frequent lifting/carrying limits (in kg), plus push/pull forces (in Newtons/kg). • Postural Tolerances: Continuous sitting and standing limits (measured in minute intervals before micro-breaks are required), and total workday postural restriction checkboxes (e.g., No Squatting, No Overhead Reaching). • Upper Limb & Joint Ranges: Specific checkboxes for forearm/wrist constraints (e.g., Avoid repetitive pronation) and grip strength levels. 2. Live Inventory Polling & Three-Tier Quoting Engine Once a consultant inputs the FCE variables, your backend pipeline must instantly check a cached live-inventory database and programmatically generate three separate, independent itemised quote packages: 1. Quote 1: Value Tier (Prioritising the lowest total cost across major networks). 2. Quote 2: Rapid Fulfilment Tier (Prioritising immediate local stock availability and the fastest dispatch times). 3. Quote 3: Premium Commercial Tier (Prioritising specialized commercial-grade gear, high adjustability, and extended warranties). To power this, you will need to build lightweight data scrapers or integrate product APIs to continuously poll pricing (in AUD), local stock levels, and delivery metrics across two distinct Australian retail streams: • Mass Retail & Big Chains: Officeworks, Harvey Norman Commercial, and Bunnings Trade. • Specialised Ergonomic Stores: Ergonomic Office, Backcare & Seating, Bad Backs, and Zenith Ergonomics. 3. AI Justification & Automated PDF Reports An integrated LLM API component (such as OpenAI or Anthropic) must ingest the structured, anonymous physical restrictions from the wizard and automatically draft formal, clinical-legal "Functional Justifications" for each selected item. The system will then compile the chosen quote package, live stock validation timestamps (locked to AEST/AEDT), and the AI-generated text into a highly professional, audit-ready PDF report template compliant with Comcare, icare NSW, WorkSafe Victoria, and WorkCover QLD guidelines. 4. Australian Data Privacy, Sovereignty & SaaS Infrastructure • Hosting: The application database, user infrastructure, and data handling must be hosted strictly within an Australian data centre region (e.g., AWS Sydney / Azure Australia East). • Privacy by Design (Anonymised-by-Default): To strictly adhere to the Australian Privacy Principles (APPs), the AI engine must only process anonymous restriction metrics. If a consultant inputs identifying details (such as a worker's real name or claim number), this data must remain strictly isolated on the client side (browser session) or masked before any payload hits the external LLM API. The final exported PDF will stitch these local details back onto the document. • B2B Subscriptions: The application is a subscription portal for corporate rehab provider firms. You will integrate Stripe Australia to handle automated recurring subscription tier billing, user account seat provisioning, and automated GST (10%) calculations. You must implement a usage-tracking middleware to monitor monthly report limits per workspace tier and prompt a Stripe top-up modal for extra usage. 5. Phased Milestone & Payment Schedule We will structure this project across 4 distinct fixed-price milestones based entirely on verifiable deliverables: • Milestone 1: Architecture & UI Setup (20% of budget) o Database schema designed and user/corporate access management deployed on local AU staging environment. o Complete frontend construction of the FCE Restriction Wizard using the specified input controls. o Stripe subscription checkout integration fully functioning in sandbox mode. • Milestone 2: Inventory & Quotes Core (30% of budget) o Background inventory polling workers active, successfully pulling live price and stock metrics from the specified mass retail and specialized ergonomic websites. o Code algorithm successfully sorting and organizing cached inventory data into three independent packages (Value, Speed, Premium). • Milestone 3: AI Engine & PDF Automation (30% of budget) o LLM API pipeline operational, cleanly mapping anonymized inputs to context-aware clinical justifications. o PDF rendering engine compiled and successfully outputting polished, formatted Insurer Justification Reports locally. • Milestone 4: Production Handover & UAT (20% of budget) o Production launch on local Australian cloud hosting. o Successful end-to-end User Acceptance Testing (UAT) loop: Input FCE → Live Refresh → 3 Quotes Generated → Stripe Account Usage Logged → Clean PDF Export. o Full handoff of cleanly commented and documented repository code.