Alexandru Ont

Machine Learning Engineer

I build computer vision and ML systems that run in the real world, from YOLO object detection deployed on NVIDIA Jetson to LLM-powered applications.

View My Work Download CV

About Me

Alexandru Ont

Hello! I'm a Machine Learning Engineer

Hi, I'm Alexandru, a Machine Learning Engineer with a Computer Science background from the Technical University of Cluj-Napoca. My focus is computer vision: for my thesis I built PARCAT, an ANPR parking-enforcement system running YOLOv8 on an NVIDIA Jetson — trained, optimized and deployed end-to-end.

I work across the full ML lifecycle — training, optimization and deployment, including on NVIDIA Jetson devices, backend APIs with FastAPI and containerized infrastructure with Docker. Lately I've also been building LLM-powered applications: RAG pipelines, vector search and agent-based systems.

Outside of work, I keep learning through research papers and online courses and enjoy turning what I learn into side projects that make everyday life smarter and simpler.

Skills & Expertise

Machine Learning & Vision

Python PyTorch TensorFlow & Keras YOLO & Ultralytics OpenCV Scikit-learn NVIDIA Jetson

LLMs & Data

RAG & Vector Search Gemini API FAISS Pandas & NumPy XGBoost & LightGBM PostgreSQL MySQL

MLOps & Backend

Docker FastAPI Flask Git & GitHub Actions Linux Node.js Jupyter

Featured Projects

PARCAT — ANPR parking enforcement system
MACHINE LEARNING

PARCAT — ANPR Parking Enforcement

Bachelor's thesis: an end-to-end license plate recognition system for parking enforcement, running YOLOv8 on NVIDIA Jetson at 92.3% mAP and 45 FPS. Full stack with FastAPI microservices, PostgreSQL and a React + Leaflet dispatcher map. Presented at the Computer Science Students Conference 2026.

YOLOv8 FastAPI PostgreSQL Docker React
Ask me about it
Football analytics with LLM agents
LLM AGENTS

Football Analytics Agents

Hackathon project built in 48h for FC Universitatea Cluj: an agent-based scouting and player-statistics system over StatsBomb event data, using PydanticAI agents with Gemini and ChromaDB for retrieval. Built with a team of 3 at "U" Hack: Code in Black & White.

PydanticAI Gemini ChromaDB StatsBomb
Ask me about it
Real-time Detection System
COMPUTER VISION

Online YOLO Detection

Web-based real-time object detection system using YOLO algorithms. Features batch analysis capabilities and confidence-based filtering with detailed detection reports and performance metrics.

YOLO Flask PyTorch
Read more
Robot Control Interface
ROBOTICS

AutoRos Robot Control

Real-time web interface for controlling a 6-axis robotic arm. Features WebSocket communication, TCP hardware integration, live camera simulation and responsive controls with bidirectional server communication.

Node.js Socket.io TCP HTML & CSS
Read more
Graph Traversal Algorithm visualization
ALGORITHMS

Graph Traversal Algorithm

Competition-winning graph traversal solution using Depth-First Search with pruning and heuristic optimization. Features cycle detection, time constraints and artifact tracking for maximum scoring paths.

DFS Branch & Bound Heuristics Python
Read more
Learning Management System
WEB APP

Learnify

A comprehensive Learning Management System for Computer Science education. Features role-based dashboards, course creation, PDF delivery, quiz management and result tracking with full CRUD operations and responsive design.

PHP MySQL HTML & CSS JavaScript
Read more

Get In Touch

I'm always open to discussing new opportunities, interesting projects or just having a chat about tech!

Email

alex.ont@icloud.com

LinkedIn

alexandru-mihai-ont

GitHub

alexandruont
Download My CV