Portfolio & Engineering Showcase
Engineering reliable software & practical ML systems.
Hi, I'm Daniel Leone. I build production-grade backend services, agentic AI tooling (Model Context Protocol), and edge computer vision pipelines with rigorous test coverage, clean system design, and zero bloat.
Featured Engineering Projects
Live DemonstrationsComputer Vision • Edge AIv0.1.0
Sprite YOLO Object Detection
Custom YOLO-mini CNN detector paired with an order-independent procedural synthetic dataset generator. 389k parameters running client-side in WebAssembly with zero server dependencies.
- • 100% mathematical parity against coursework baseline (0.941 mAP)
- • Zero git bloat: eliminated 280+ MB static images
- • 52 automated tests in GitHub Actions CI
Backend API • Agentic Tooling Live on Railway
Inventory Audit API & MCP Server
Production FastAPI service managing warehouse cycle counts, inventory audits, and variance analytics with PostgreSQL 16 Row Level Security, plus a dedicated Model Context Protocol (MCP) server for AI pair programmers.
- • Pure functional rule engine (
decide_audit) - • SQL window functions for cumulative variance reporting
- • Fail-closed authentication, IP rate-limiting, 150+ tests (99.8% coverage)
Core Competencies & Toolchain
Languages & Runtimes
Python 3.12
SQL (PostgreSQL 16)
TypeScript / JavaScript
Node.js • WebAssembly
Backend & APIs
FastAPI • Pydantic v2
SQLAlchemy 2 • Alembic
Model Context Protocol (MCP)
Row Level Security (RLS)
Machine Learning & CV
TensorFlow • Keras 3
YOLO Architecture & Loss
ONNX Runtime • TFLite
Procedural Synthesis
DevOps & Cloud
Docker & Compose
GitHub Actions CI/CD
Railway • Supabase
Cloudflare Pages & Proxy