Port Local Python CDS Prototype to Modern B2B Web App (FastAPI + React/Next.js + LLM)

We have a validated, functioning prototype for a clinical nutrition decision support platform developed locally using Python and Claude Code. The prototype captures patient biometrics, evaluates pathology constraints (e.g., T2D, dyslipidemia, hypertension), executes deterministic macro calculations against institutional food databases (INTA Chile / USDA), and formats personalized meal plans via structured LLM workflows. We need to port this working local codebase into a clean MVP platform to test it in the market. We are looking for an MVP that has a cloud-hosted B2B web application with practitioner authentication, a clinical dashboard, and a clean, high-resolution PDF export engine. What We Have Working Local Prototype: Complete Python scripts executing all clinical formulas, database matching, and structured LLM prompts. Structured Datasets: Clean institutional food databases and clinical pathology rulesets already configured. Clear UI/UX Wireframes: Defined screen states and practitioner workflows. Code walkthrough provided to shortlisted candidates. Core Deliverables (Scope of Work) Backend & AI Pipeline (FastAPI / Python): Port existing local Python scripts into a modular FastAPI backend. Connect existing deterministic calculation logic with the LLM layer (Anthropic Claude / OpenAI API). Crucial: Clinical math and food-database portion constraints are handled deterministically in Python; the LLM layer is strictly used for structured JSON recipe synthesis, culinary formatting, and patient-specific rationales. Enforce strict JSON Schema / Pydantic models on all LLM responses. Practitioner Web Dashboard (Next.js / React + Tailwind CSS / shadcn/ui): Intake Screen: Rapid capture of patient biometrics, pathology selections, and clinical targets. Target Review: Interactive sliders for clinicians to review/override calculated calories and macros. Plan Canvas: Interactive view displaying meal structures, allowing meal regeneration and ingredient swaps. Clinic Profile: Basic settings to store practitioner credentials and clinic branding (logo, colors). Practitioner Auth & Secure Database: Implement secure practitioner authentication (Email/Password + Google OAuth via Supabase Auth or similar). Provision PostgreSQL schema to manage doctor profiles, patient intake history, and saved plans with tenant data isolation. Dynamic Branded PDF Engine: High-resolution, multi-page PDF generation (e.g., via @react-pdf/renderer or Python WeasyPrint). Renders the doctor’s uploaded clinic branding alongside structured, AI-personalized meal schedules, preparation instructions, and clinical dietary explanations. Must cleanly manage page breaks and print stylesheets without visual clipping. Cloud Deployment & Handover: Containerize and deploy the application to cloud infrastructure (Vercel for frontend, Render/Railway/AWS for backend, Supabase for PostgreSQL). Secure environment variables and provide setup documentation for clean code handoff. Required Qualifications Proven track record with Python (FastAPI) and modern React / Next.js (TypeScript). Production experience integrating LLM APIs using structured JSON output modes (Pydantic / JSON Schema). Demonstrated experience building dynamic, multi-page PDF generation pipelines from database/API data. Experience with PostgreSQL schema modeling and standard SaaS authentication. Ability to use modern UI component libraries (shadcn/ui, Tailwind CSS) to build clean enterprise interfaces fast without dedicated Figma designers. Proposal Submission Requirements Please submit: Total Fixed-Price Quote broken down by milestone (Backend/API, Frontend Dashboard, PDF Engine, Deployment). Estimated Turnaround Time in weeks. Recommended Tech Stack (specifically for your proposed PDF generation and hosting architecture). Relevant Work Samples: 2 links or code repositories showing similar full-stack dashboards, LLM structured implementations, or automated document exports you have built.