Marius Vlas

This project revolved around a web app for one of the largest airline companies in the US. I was the sole front-end developer and closely worked with the back-end and data-science teams to develop the most efficient way to display fueling data in a meaningful way for data scientists and higher management alike.

The app has been used by the data-science, accounting, and management teams. Its primary role was to present vast amounts of tightly coupled flight data in significant visual representations.

My responsibilities:
• Conceptualized and designed the app and its features.
• Developed the entire front end.
• Facilitated efficient and transparent communication with all actors involved (the back-end team, the DevOps team, and the data-science team).

The project ended up being a success—significantly optimizing the internal data-analysis processes of my client.

This project revolved around a web-based order-management app for a German-US-based construction and waste-disposal startup. I served as manager for a team of five developers and two QA testers.

My responsibilities:
• Created and maintained the front-end architecture at the highest standard.
• Managed, guided, and assured timely deliveries by front-end and QA teams.
• Orchestrated and facilitated communication between client, product, design, development, and QA teams.

Alt Parts was a Swiss-bootstrapped B2C intelligent search engine built for car enthusiasts worldwide who are tired of wasting time hunting for the right parts. As the product and technical founder, I owned the entire product and engineering surface area end-to-end: roadmap and specs, UI/UX, front-end architecture, back-end services, data flows, AWS infrastructure, CI/CD, and production operations.

I built the React web app and the Node.js/Python services that aggregated marketplace data and ranked results against a user’s vehicle. On the “intelligence” side, I implemented an LLM/NLP layer for query understanding and refinement, and I personally trained and deployed a domain-specific NLP/NER service (Python, spaCy) to extract automotive part entities and fitment signals from messy real-world queries. To drive iteration with real evidence, I instrumented GA and GTM event tracking and built BigQuery reporting to analyze funnels and user behavior, feeding those insights directly into product decisions and roadmap updates.

We were selected for a booth at Web Summit Lisbon, and our pitch was showcased on the main entrance display on day one. At peak, the product reached around 30,000 daily users.