Hi, I'm Lorenzo SignorelliLorenzo Signorelli

Full-Stack Developer & AI Enthusiast

I build scalable web apps and AI-powered products. Based in Italy, I specialize in React, Node.js, Python, and Machine Learning. Let's create something impactful together!

ReactTypeScriptNode.jsPythonNext.jsAI/ML
Lorenzo Signorelli
I don’t use libraries.
I adopt problems.

What I Build

Full-stack engineering with a focus on performance, scalability, and user experience.

Frontend Engineering

Modern React, Next.js, TypeScript. Pixel-perfect UI with Framer Motion.

Backend & APIs

Node.js, Python, serverless. Scalable APIs and database design.

AI & Machine Learning

LLMs, computer vision, NLP. Building intelligent products.

DevOps & Cloud

Docker, CI/CD, Vercel, AWS. Ship fast, scale reliably.

How I Work

A structured, outcome-driven process that takes products from strategy to production.

01
01

Discover

Align goals, stakeholders, and constraints to define requirements and clear success criteria.

Next →
02
02

Build

Design and build robust solutions with clean architecture, best practices, and high code quality.

Next →
03
03

Iterate

Validate through testing, feedback, and metrics; continuously improve UX, performance, and reliability.

Next →
04
04

Ship

Release to production with CI/CD, monitoring, documentation, and ongoing support.

Launch ✦

Featured Work

A selection of my most impactful projects, from AI research to production applications.

Git RAG Assistant

Git RAG Assistant

Local codebase Q&A with RAG: fast retrieval, grounded answers, and a simple chat UI.

  • Runs entirely locally (repos + embeddings + retrieval)
  • Source-grounded answers with citations/snippets
  • Next.js UI + FastAPI backend
RAGLLMFastAPI+2
Basket-AI

Basket-AI

Affordable basketball tracking R&D: YOLO-based detections, jersey OCR, and MOT experiments.

  • Player and ball detection with custom YOLO models
  • SAM2 segmentation + color clustering
  • Jersey number localization + PARSeq OCR pipeline
Computer VisionYOLOOCR+2
Instance-Guided UDA

Instance-Guided UDA

Refining pseudo-labels in Unsupervised Domain Adaptation for semantic segmentation using Segment Anything Model (SAM2).

  • Pseudo-label refinement with SAM2
  • Two prompting strategies: Informed & Grid-based
  • Evaluation on ScanNet dataset
Computer VisionDeep LearningRobotics+3
0+

Products Shipped

0+

Years Crafting Software

0x

Leadership Engagements

0+

Technologies Mastered