Riccardo Scrizzi

I built an AI-driven system that turns a marketer's informal brief into brand-compliant, multilingual, Klaviyo-ready HTML emails for an eCommerce brand. It reverse-engineers the client's brand—typography, color rhythm, CTA style, layout rules—into a structured brand guide and a modular block library, then uses Claude (Anthropic) to generate on-brand emails across 6 languages with correct localized URLs and Outlook-safe, table-based HTML. It lets a non-technical operator produce production-ready campaigns from a brief, replacing ad hoc manual design. I designed the brand-guide system, the composable template formulas, the translation pipeline, and the Claude-driven generation workflow.

I designed and built a media discovery web app backed by a catalog of 1.4M+ titles. The React SPA delivers a polished browsing/detail experience (search, ratings, reviews, themed shelves); the Node.js/Fastify + MongoDB back end serves it from a dataset I assembled by aggregating and reconciling multiple sources (TMDB, TVDB, critic-score providers) through a nightly automated sync pipeline, with weighted aggregate scoring and cross-source conflict-resolution rules. Built solo, end-to-end, AI-augmented, on self-managed Linux infra (Caddy, PM2, CI/CD).

I built a Three.js 3D configurator for a company selling premium shower systems, which sales reps use on tablets during in-home consultations. Using the tablet camera, the app placed the shower in the client's real room (AR-style), with real-time changes to colors, materials, and finishes. I developed the full 3D rendering pipeline—physically based materials, camera handling, and the configuration UI —letting reps deliver personalized demos and capture custom orders on-site.

I built a highly interactive web dashboard with Konva.js for an industrial wine producer, letting operators visualize, monitor, and control machinery in real time—water filtration, boiling, and other production-critical processes. It communicated with industrial hardware via REST APIs to trigger and track the live machinery state, with custom SVG controls and safety-alert mechanisms. I owned both the UI/UX and the architecture, ensuring the interface reflected machinery state and commands with zero tolerance for error.

I built a full-stack interactive geospatial platform that helps digital nomads and investors compare cities worldwide on cost of living, real estate, and lifestyle. The React + React-Leaflet front end lets users simulate monthly expenses via a personalized "basket" and compare regions through map overlays and data tables (housing/rent, climate, pollution, safety, quality of life, salary). On the back end (Node.js), I built large-scale scraping pipelines with proxy rotation and high concurrency, aggregating data from many public and private sources into a clean, queryable dataset.