Skip to main content
All work

Own product · 2025

Kreditožrouti.

A course scheduler for VŠE students, live at kreditozrouti.cz. I created it and run it alone, from the InSIS scraping to the deployments. The numbers are modest on purpose: it quietly saves a few students their semester, every day.

Role
Creator & Sole Maintainer
Timeline
Sep 2025 – present
Sector
EdTech · Higher Education
Kreditožrouti in dark mode: the first step of the study plan wizard, choosing a faculty
Students
200+
Finished the setup wizard
Onboarding
35%
Of 589 visitors finish setup
Status
Live
Monitored, auto-deployed

The challenge

Kreditožrouti is a course scheduler for students at the Prague University of Economics and Business. It scrapes public InSIS course data and puts a faster, clearer layer of search and scheduling on top, so a student can close a gap in their timetable in seconds instead of clicking through the university system.

It was never a startup, and the metrics are modest on purpose: a few hundred students, a handful of them every day, found mostly through Discord, Facebook and word of mouth. Alongside my client work, it is the one product I own end to end: live, maintained and still gaining users.

What I owned

  • InSIS scraping and scheduling

    Scrapes public InSIS course data and puts better search and scheduling on top of it than the university system offers, so a student can fill a gap in their timetable in seconds. The name is a joke that means something: when a timetable is set badly, courses eat your ECTS credits. Credit-gobblers.

  • From hackathon to product

    Born at the 4FIS Mini Hackathon on live scraped data, a working MVP in 3.5 hours, first place. When the course that later hosted it ended, I picked it up and carried it alone to a finished product rather than a demo.

  • Production infrastructure

    Runs today with monitoring, alerting and automatic deployments. Vue on the front, Node with Redis and BullMQ workers behind it, containerized with Docker and Traefik.

  • Bachelor thesis

    The InSIS work led directly to my bachelor thesis: an MCP server over InSIS data, hosted on university servers under an official domain.

Architecture

The browser loads the Vue app and calls the API through Traefik. The API reads and writes MySQL and queues scrape jobs in Redis. A scraper worker takes jobs from Redis, fetches InSIS course pages and returns results through Redis. AI clients reach a separate MCP server through Traefik; it only reads MySQL.

Observability: Alloy Prometheus and Loki Grafana. Alertmanager Discord. Umami for analytics.

A pnpm and Turborepo monorepo. The API owns every database write. The scraper only fetches and parses InSIS, then hands results back over a queue, so a slow university server never blocks a student's search. The MCP server reads the same MySQL data for AI clients and my bachelor thesis.

Stack

  • TypeScript
  • Vue.js
  • Node.js
  • Redis
  • BullMQ
  • Docker
  • Traefik
  • CI/CD
  • Git
Kreditožrouti in light mode: the first step of the study plan wizard, choosing a faculty
Kreditožrouti week timetable with seven courses and one conflict, lecturer names hidden
Kreditožrouti course list with faculty, ECTS and schedule columns next to the filters
Umami dashboard of Kreditožrouti: 589 visitors, 783 visits, 2.14k views, 13% bounce rate and monthly bars from June to October

Let's take yours to production.

Part-time B2B, around 10 hours a week. Hybrid in Prague or fully remote. I reply within 24 hours on weekdays.

contact --jan

call
30 min, free
reply
< 24 h on weekdays
status
open to part-time B2B