I'm Alessandro Magnocavallo, a developer with a background in Communication Design from Politecnico di Milano. I focus on bridging design thinking with technical implementation, specializing in AI integration and prompt engineering.

A web-based music player with an iTunes-inspired interface, featuring intelligent playlist generation and track recommendations.

AI Implementation:

• Playlist Generation: Sends complete library metadata to Claude API with structured prompts that analyze musical coherence, genre compatibility, temporal cohesion, and energy characteristics. The AI creates thematic playlists while avoiding album-centric grouping.
• AI Next Song: Predictive track selection that preloads recommendations 20 seconds before the current track ends. Uses weighted sampling (50% genre match, 30% decade match, 20% random) to optimize API token usage for large libraries.
• Caching Strategy: Playlists cached for 7 days with MD5 hash invalidation on library changes. Next song selections cached 24 hours with 60-75% expected hit rate.
• Fallback System: Deterministic algorithms activate when API is unavailable, maintaining functionality without AI.

A static photography portfolio with client-side semantic search powered by AI-generated metadata.

AI Implementation:

• Automated Tagging Pipeline: Python script (tag_photos.py) processes the image library through Vision AI models, generating structured JSON metadata for each photograph.
• Extracted Taxonomies: Photography style and technique, lighting conditions, scene context, object detection with confidence scores, and dominant color palette (HEX codes with coverage percentages).
• Semantic Search: Client-side search indexes all metadata on page load, supporting both structured queries (style:street, lighting:golden hour) and free-text search against AI-generated descriptions.

A Telegram bot for automated financial analysis and portfolio management, deployed on Cloudflare Workers.

AI Implementation:

• Multi-Step Analysis: The /analyze command triggers Claude 3.5 Sonnet to collect market data via Tool Use, then generates structured reports covering fundamental analysis, technical analysis, and weighted scoring (Technical 35%, Fundamentals 30%, Sector 20%, Sentiment 15%).
• Natural Language Parsing: Accepts both structured commands (/buy AAPL 150 10) and conversational input ("I bought 10 Apple shares at 150 dollars"). AI parses intent when pattern matching fails.
• Durable Objects: Background processing via Cloudflare Alarms handles long-running AI executions without timeout issues.

Minimal landing page hosted on GitHub Pages at a.magno.me.

Provider	Certification
	UX Design
	AI Fluency: Framework & Foundations
	Building with the Claude API
	Claude Code in Action

Website

University Portfolio

[email protected]