---
title: "studenti.rs: a student platform, eighteen years running"
description: "A Serbian student community and document library online since 2008, bought back in 2022, rebuilt and monetised, with operations run largely by agents."
canonical: "https://mladenruzicic.com/work/studenti-rs"
last_modified: "2026-11-16"
author: "Mladen Ružičić"
language: "en"
---

# studenti.rs

> Serbian student platform, running since 2008. Built it at my first job, bought it back in 2022.

By [Mladen Ružičić](https://mladenruzicic.com). A long-running Serbian student community and document library. I built it at Nemestic Studio, my first job in tech, and acquired the platform from them in 2022, then rebuilt it, monetised it with PRO subscriptions, and now run its moderation, payments and operations almost entirely on automation.

- **Status**: Running
- **When**: 2008 to now
- **Role**: Original developer, later owner and operator
- **Website**: https://studenti.rs

## In numbers

- 250k: registered students
- 40k: documents
- 2008: running continuously since

## Summary

studenti.rs is a Serbian student platform I built at my first job. It kept running. In 2022 I bought it back, rebuilt it, and turned it into a small business I operate almost entirely through automation. It is the longest feedback loop I have on any decision I have ever made.

## Role and scope

Original developer, while working at [Nemestic Studio](/about), my first job in tech. Owner and sole operator since 2022, when I acquired the platform from them: product, rebuild, infrastructure, moderation policy, monetisation, support and the agent systems that run the day-to-day.

## Context

Students need material: notes, past exams, documents. The platform accumulated it for well over a decade, along with the search visibility that comes from being the thing that was there first and never went away. That audience is the asset. It is also a moderation liability, because a user-generated document library attracts exactly what you would expect.

## Problem

An old platform carries two debts. The technical one is obvious and boring. The operational one is the real constraint: user-generated content needs review, payments need handling, reports need answering, and a single owner with a full-time job cannot be the queue. Before 2022 the site survived on neglect, which caps what it can be.

## Decisions

Buying it back was the first decision, and the reasoning was simple: the audience and the archive were real and irreplaceable, and the code was not. So the code got replaced.

Monetisation went to PRO subscriptions rather than to more advertising. Ads on a student site pay badly and make the product worse for the people using it most.

Operations were automated rather than staffed. Content approval, triage, scheduling, payments, reimbursements and report handling run through an agent system on infrastructure split between Cloudflare and self-hosted machines at Hetzner. An internal RAG layer with hybrid search gives it the platform's own history to answer from, so decisions are consistent with precedent rather than reinvented each time.

Automation runs on policy, not on vibes. The rules for what is acceptable are written down, the agent applies them, and edge cases escalate to me. Automating a judgment you have not made yourself only produces a faster wrong answer.

## Contribution

Built the original platform at my first job. Bought it, rebuilt it and now run it: architecture, infrastructure, subscriptions, moderation policy, and the operational agents that handle approvals, payments, infrastructure health, dependency updates, bug fixes and analytics.

## Outcome

Eighteen years live, a quarter of a million registered users, tens of thousands of documents, and a subscription product, all operated part-time by one person. All of its traffic is organic. Revenue figures stay private.

## Learnings

Longevity is a strategy. A product that simply keeps working for a decade accumulates an audience and a search position that money cannot easily buy.

Operational load, not feature work, is what kills solo products. Every hour of manual review is an hour the product cannot grow, which makes automating operations a product decision rather than an engineering indulgence.

## On the page

A student community in Serbia, online since 2008. I built it at my first job, bought it back in 2022, and rebuilt it this year.

Running · 2008 to now · Original developer, then owner and sole developer

- **250k** Registered students
- **40k** Documents
- **2008** Online since

studenti.rs is where students in Serbia share study material: notes, past exams and papers. All of its traffic is organic.

I bought it back in January 2022. The audience and the archive were real. The code was not, and the site had survived on neglect.

I completely redesigned it this year and automated the old stack. Moderation, payments and reports run on written rules, and only the edge cases reach me.

Written about it:

- [15 Years Later, I Bought Back My First Website](https://medium.com/@ruzicic/15-years-later-i-bought-back-my-first-website-and-now-its-my-most-interesting-project-yet-e17d034df2f7), Medium
- [14-fazni document pipeline](https://studenti.rs/blog/upload-pipeline/), studenti.rs blog, in Serbian

### The pictures

- The site today: searching, then opening a document. (Clip: Searching studenti.rs from the home page, then opening a document at its free preview.)
- 2010, the earliest copy in the Web Archive. (Picture: studenti.rs in April 2010: a plain white page with the logo beside black silhouettes of students, a grey menu bar, a list of new papers and a list of categories.)
- The whole loop. The next squares show each part. (Diagram: A loop in five steps: a student uploads notes, it is checked and sorted, it becomes a page, search finds it, and another student opens it, which leads back to the first step. The library grows and stays clean without anyone reading uploads by hand. What still reaches me: uploads the final check is unsure about, held until I look; lifting a flag from an account; an alert when an upload gets stuck.)
- What happens to one upload, and why each check is there. (Diagram: A flowchart in three lanes: the student, the system and me. The student uploads a file. A model asks if it is study material; if not, it is refused on the spot. The student confirms the details. An automatic check asks if it is a copy; if so, it is binned, and a strike. A model asks if it is good enough: if not, it is binned; if it is not sure, I decide; if yes, it is published now or on a future date. Published, the student is rewarded and told.)
- Any file a student uploads becomes a readable web page on its own, and its text feeds the site’s search. (Diagram: A pipeline. A student uploads any office file: Word, PowerPoint, PDF or another format. It is converted to a PDF, its words are extracted as Markdown, and a web page is published. One branch sends the Markdown to the site’s search index. All 45,000+ older documents were re-run. 7,000+ were removed: duplicates, low quality, or material that should never have been public.)
- What search could see before the rebuild, and what it sees now. (Diagram: Search, before and now. Before the rebuild, search matched titles and page text; the rest of a document was not searched. Now search reads whole documents and understands meaning, so a query can find a passage that says the same thing in other words. Before, 1 in 10 searches ended with the student finding what they wanted; 7 in 10 do today.)
- In five days 20,773 documents were linked to their institution; the rest were left blank instead of wrongly filled. (Diagram: A count. The old library was a mess: badly written titles, missing tags, a catch-all subject and documents under the wrong institution. Then 38,263 documents were sorted by institution in January 2026. 20,672 matched from the text, 101 by reading the whole PDF, 57 still ambiguous, 15,367 with no institution named in the text, 2,066 with none confirmed from the PDF.)
- About four in ten documents were filed under the wrong institution. Now none are. (Diagram: Before and after. Before the clean-up about 40% of documents were filed under the wrong institution; on 21 January 2026 none were, after 15,237 wrong links were removed. Filed under the right one: 35.7% before, 54% after. The other 46% never name an institution in their text and are left blank, not guessed.)
- A document page: what a student comes for. (Picture: A studenti.rs document page: a seminar paper on motor vehicle brakes, its first page beside a panel with the reading time, the page count, a summary, tags and the institutions it belongs to.)
- A Grok bot answers questions and reviews payments every day, without me. (Diagram: A flowchart in three lanes: the student, the Grok bot and me. Every day a student writes to info@: questions, payment reviews. The bot reads and sorts the mail and asks if it is simple. If yes, it checks the facts and replies in the site’s own voice, and the answer is sent without me. If not, I answer what it hands me: refunds, goodwill, billing disputes and legal.)
- The home page on a phone. (Clip: Scrolling down the studenti.rs home page on a phone, from the green header with search.)

## Images

![Searching studenti.rs from the home page, then opening a document at its free preview.](https://mladenruzicic.com/static/work/studenti-rs/square-clip-search-open-poster.webp)
![studenti.rs in April 2010: a plain white page with the logo beside black silhouettes of students, a grey menu bar, a list of new papers and a list of categories.](https://mladenruzicic.com/static/work/studenti-rs/sq-home-2010.webp)
![A studenti.rs document page: a seminar paper on motor vehicle brakes, its first page beside a panel with the reading time, the page count, a summary, tags and the institutions it belongs to.](https://mladenruzicic.com/static/work/studenti-rs/sq-document.webp)
![Scrolling down the studenti.rs home page on a phone, from the green header with search.](https://mladenruzicic.com/static/work/studenti-rs/phone-home-scroll-poster.webp)

## Links

- [studenti.rs](https://studenti.rs)

## More case studies

- [TenderLift](https://mladenruzicic.com/work/tenderlift): Swiss public-procurement intelligence for supplier-side SMEs: find, assess and win relevant tenders.
- [FontAlternatives](https://mladenruzicic.com/work/fontalternatives): Find free alternatives to expensive typefaces. An experiment that grew into steady organic traffic.
- [Zarez](https://mladenruzicic.com/work/zarez): A spelling checker for Serbian, in Latin and Cyrillic.
- [Panciona](https://mladenruzicic.com/work/panciona): A white-label app for coaches, trainers and nutritionists to give to their own clients, on their own domain.
- [ZF SCALAR](https://mladenruzicic.com/work/zf-scalar): From individual contributor to engineering manager of the platform team, to product owner of several products.
- [Shopify](https://mladenruzicic.com/work/shopify): Shop Minis from zero: mini React Native apps inside Shop.app, plus the SDK and partner ecosystem.
- [HEGIAS](https://mladenruzicic.com/work/hegias): Moving an AR/VR MVP toward production: 3D pipelines, microservices on AWS, in-app voice.
- [WolkAbout](https://mladenruzicic.com/work/wolkabout): Junior to senior in three years: Angular frontends for an IoT platform moving large volumes of live data.
- [Work he has put real time into](https://mladenruzicic.com/work)
