Michele Marcucci

Hi, I'm Mike.

AI Software Engineer.

Full-stack developer in Milan, wiring AI into real products. I build AI pipelines., desktop apps., web platforms.

Contact me
See projects
Illustration of Michele Marcucci holding a mug

Selected work

From native macOS apps to AI tooling and computer vision. More on GitHub.

  • Alembic

    Visit site

    An enterprise platform for RAG, document sharing and management, powered by MCP tooling and local AI. Privacy-first and local-first, built for individuals and companies.

    • RAG
    • MCP
    • Local AI
    • Privacy-first
  • GymTonic workouts tab: two of three workouts done this week and the list of routines
    Rest timer with 54 seconds left and the heart rate read from a chest strap
    A set in progress: 10 reps at 45 kg, with last session's weight as a reminder

    GymTonic

    Coming to Google Play

    An Android gym diary: write your routines and run them set by set, with a rest timer that keeps going with the screen off. Offline, no account, no ads.

    • Kotlin
    • Jetpack Compose
    • Android
    • Offline-first
  • OneClick menubar panel with toggles for hidden files, dark mode and caffeinate

    OneClick

    View code

    A native macOS menubar utility that puts quick system actions one click away. My most starred project.

    • Swift
    • SwiftUI
    • macOS
  • Original photo of a dirt path through a rapeseed field

    NPR Painting

    View code

    A non-photorealistic rendering filter that turns photos into oil-painting-style artwork.

    • OpenCV
    • Python
  • Terminal running Skin Detection on a video with eight threads

    Skin Detection

    View code

    Human skin detection and segmentation from images and video, using classic computer-vision techniques.

    • OpenCV
    • Image Processing
    • Python

  • LocalAI Chatbot

    View code

    A Telegram chatbot wired to a local LLM through Ollama. Private, self-hosted conversational AI.

    • Ollama
    • Telegram
    • Python

Websites I've built

The tools I reach for

View CV

Languages

Python and TypeScript every day, Swift and C# when the platform needs it.

  • Python
  • TypeScript
  • Swift
  • C#

AI and data

Training, fine-tuning and vision work, plus local models.

  • PyTorch
  • Hugging Face
  • OpenCV
  • Ollama

Web and shipping

Interfaces for the products, containers to run them.

  • React
  • Next.js
  • Tailwind CSS
  • Docker

How I work

I design, build and optimize software end to end, and reach for AI when a problem calls for it. I studied Information Technology at the University of Milano-Bicocca.

  1. Full-stack, end to end

    Clean architecture, databases, REST APIs, frontends and Docker. I take software from the first diagram all the way to production.

    • Clean Architecture
    • REST APIs
    • Databases
    • Docker

  2. Built to optimize

    Working software is the starting point. Then come careful refactors, performance work and pipelines that scale.

    • Performance
    • Refactoring
    • CI/CD

  3. AI, put to work

    RAG systems, data extraction and processing pipelines that turn messy, unstructured data into something useful.

    • RAG
    • Data Extraction
    • LLMs

  4. Specialized models

    When off-the-shelf models fall short, I fine-tune specialized LLMs on curated datasets for domain-specific precision.

    • Fine-tuning
    • Datasets
    • Hugging Face

Have a project in mind?

Write me an email or find me on LinkedIn.

michymarcucci@gmail.com