AI Platform Development - Web & App version

Budget: Monthly $2000 - $2500 TimeZone: EST | Online Time: 9 am - 5 pm Communication: Native English Speaking Skill Preferred Option: 2-3 mins of loom video for brief introduction yourself 1. Data Ingestion & Intelligence Foundation - Automated ETL Pipelines: Design and build modular data ingestion pipelines to collect, parse, and synchronize catalog data from multiple external retailers, APIs, and affiliate feeds. - Centralized Catalog & Extensible Taxonomy: Architect a scalable relational database structure with a flexible, hierarchical taxonomy framework capable of supporting future category expansions. - Entity Resolution & Normalization: Develop robust normalization logic to identify, deduplicate, and consolidate records across multiple data sources (e.g., name, brand, size, and universal identifier matching). - Structured Attribute Mapping: Implement a dynamic attribute tagging system to categorize complex product features, constraints, and metadata for downstream processing. 2. AI-Powered Recommendation Engine - Multi-Factor Scoring Algorithm: Architect a deterministic scoring model that evaluates and ranks items based on user profile compatibility, attribute alignment, budget constraints, and category relevance. - Dynamic User Profiling: Build a comprehensive user preference system to capture, store, and update granular user constraints, favorites, and avoidance criteria. - Conversational AI Integration: Develop an LLM-powered assistant interface to handle natural language queries, product comparisons, routine/guided discovery, and contextual Q&A. - Explainable AI (XAI) Layer: Implement a reasoning layer that generates transparent, logic-backed explanations for why specific items are recommended, ensuring outputs are grounded in the scoring engine rather than independent LLM generation. 3. User Experience & Workflow - Lightweight Session/Cart Management: Build functionality for users to save items, curate custom lists, revisit historical recommendations, and seamlessly redirect to external partner checkout flows (no native payment processing). - Responsive Frontend Architecture: Develop a mobile-first, fully responsive web application ensuring optimal performance and UI consistency across desktop, tablet, and mobile devices. 4. Platform Infrastructure & Administration - Secure Authentication: Implement robust user registration, login, password recovery, and session management (including OAuth support). - Admin & Operations Dashboard: Develop internal tools for catalog oversight, user management, recommendation quality monitoring, and AI token/usage cost tracking. - DevOps & Handoff: Establish automated CI/CD deployment pipelines, environment configuration, and deliver comprehensive technical documentation and source code handoff. 5. Technical Stack - Frontend: Next.js 14 (React) + Tailwind CSS (for SSR, SEO, fast UI, and mobile responsiveness). - Backend / API: Node.js + Express (or Next.js API Routes) for lightweight, AI-friendly architecture. - Database: PostgreSQL (via Supabase) for relational product/profile data + Redis cache to optimize AI response costs. - AI / LLM: OpenAI API with model routing based on task complexity (GPT-4o for advanced, GPT-4o-mini for mid-tier, GPT-3.5 for low-complexity tasks). - Vision / Image Analysis: OpenAI Vision API (GPT-4o) for future look recreation, skin analysis, and image interpretation. - Authentication: Supabase Auth (or Auth.js) for out-of-the-box OAuth, email/password, and session management. - Product Ingestion: Node.js cron jobs + affiliate feed parsers for daily automated catalog updates. - Cloud / Deployment: Vercel (Frontend) + Railway or Render (Backend) for cost-efficient, scalable, low-DevOps overhead hosting. - Analytics: Custom admin dashboard + PostHog (optional) for tracking AI costs/queries, user retention, and feature usage. Whole Project Roadmap and Plan was prepared. will share it for candidate person.