Jovan Jovanović

Machine Learning & Applied AI

Profile

I’m a Computer Science student who enjoys difficult problems, thoughtful engineering, and learning by building things that actually work.

Education: BSc Computer Science, Faculty of Electronic Engineering, University of Niš 2022-2026
Focus: Embeddings, RAG, vector search, model evaluation, and data preprocessing

Selected projects

word2vec from scratch Python, NumPy Pure NumPy skip-gram with negative sampling, including the forward pass, loss, and embedding gradient updates. Cloud Latency Scraper TypeScript, Three.js, Bun An interactive 3D globe showing real-time latency and status data scraped from AWS, GCP, and Azure. EU AI Act RAG Next.js, Mastra, pgvector A grounded legal assistant that cites retrieved source passages, refuses low-confidence answers, and reached 95.1% sample accuracy.

Capabilities

Machine learning

PyTorch, Hugging Face, supervised and unsupervised learning, evaluation metrics, embeddings, and retrieval systems.

Engineering

Python, TypeScript, SQL, PostgreSQL, SQLite, Git, Docker, Linux, Next.js, and Vue.

Coursework

Artificial Intelligence, Computer Vision, Data Structures, Databases, and Object-Oriented Programming.