Atama Koçu

Overview

Atama Koçu is a KPSS preference simulator I built for candidates applying to Turkish Ministry of Health placements. It covers every title, from nurses and midwives to health technicians and medical secretaries, across bachelor's, associate and high school levels. Candidates pick their branch, enter their score and get an answer to "where can I get in with this score?" based on past placement data. They can then build a preference list and test the placement before applying. I developed the backend, the web app and the mobile app from scratch. I wrote the API with Express.js and MongoDB, and built a data pipeline that parses the official results PDFs published by ÖSYM into the database. Judging placement chances from a single cutoff score would be misleading, so I wrote a seeded Monte Carlo simulation engine that accounts for cutoff shifts between periods: the same input always gives the same result. With crowd analysis, candidates can see where they stand among others applying to the same positions, through an anonymized pool. The in-app Atama Koçu chat assistant answers from its own data first and builds a preference list step by step. It calls an LLM only when needed and only if the user's plan allows it, which keeps AI costs under control. I handled long-running jobs asynchronously with RabbitMQ queues and workers, and used Redis for caching and rate limiting and Socket.IO for real-time notifications. I built the web app with Next.js and the mobile app with React Native (Expo), both mobile-first. I set up subscriptions and plans with RevenueCat, and built an admin panel with two-factor (TOTP) authentication. The whole system runs in Docker containers.

I built Atama Koçu end to end: a preference list and placement simulator for KPSS candidates applying to Turkish Ministry of Health placements. I used Express.js, MongoDB, RabbitMQ, Redis, Next.js, React Native and Docker.