Liviu .ioNESI
package dev.portfolio;
import org.springframework.boot.SpringApplication;
import org.springframework.boot.autoconfigure.SpringBootApplication;
@SpringBootApplication
public class LegacyBackendApplication {
public static void main(String[] args) {
SpringApplication.run(LegacyBackendApplication.class, args);
System.out.println("Java EE Enterprise Logic Initialized...");
}
}Work experience
<Senior Software Developer/>
<Software Developer/>
<Junior Software Developer/>
<Internship in Software Development/>
Education
<dual study/>
State-Certified Computer Scientist— Technical Systems
<master/>
Master— Mathematics Teacher Education
<bachelor/>
Bachelor— Mathematics and Computer Science
<college/>
High School Diploma (Baccalaureate)— Mathematics and Computer Science
Projects
This portfolio site. It looks like the Antigravity IDE, with an activity bar, file explorer, search panel, status bar and command palette. It runs on the Next.js 16 App Router with the React Compiler. It has 4 languages (English, German, French, Romanian) and CV export to PDF. The contact form uses Google sign-in and Nodemailer, and shows the distance between the visitor and the office with the Geolocation API. Unit tests cover about 97% of the code, and Husky, ESLint and Prettier check every commit. The site scores 100 in all four Lighthouse categories.
The website and documentation of Brevier, a PDF library with the formula HTML + CSS + JSON = PDF. Brevier has clients for Python, C#, TypeScript and Java, and supports German business documents: DIN 5008, hyphenation, multi-page tables and e-invoices (ZUGFeRD / Factur-X). The site is built with Nuxt 4 and prerendered to static HTML. It ships no framework scripts in production, uses no cookies or tracking, and comes in English and German. It scores 100 in all four Lighthouse categories.
My answer to a software architect challenge from Kärcher: a system for 1,000 robots and 100 operators. Part 1 is the design for AWS (IoT Core, Kinesis, Valkey, EKS, Cognito, Terraform), written as an arc42 document with an ADR for each decision. Part 2 is a working prototype: an MQTT broker with TLS and a login per robot, a Python backend, a Next.js dashboard and an Android app (Capacitor). One docker compose command starts it. The demo is live with simulated robots. The source code is private.
LMDB (Live Movies Database) is a microservices platform that copies the TMDB v3 API, so the React app can use it in place of api.themoviedb.org. It saves movie and actor data on the first request and then serves it from its own database. It has 8 Spring Boot services on Java 25 with Spring Cloud, and uses MongoDB, PostgreSQL 17 and Redis. A local AI layer (Spring AI, Ollama, pgvector) gives chat, semantic search and natural-language search, and Vosk gives offline voice control. No paid AI API is used. Terraform deploys it to Azure AKS or AWS k3s on a $0 budget. It has 1,157 automated tests, 24 ADRs and CI/CD on GitHub Actions.