PlaceReady: Interview practice handouts from a job description, running locally

This is a submission for the Hacktoberfest Open-Source AI Challenge Week 2: Side Hustle

What I Built

PlaceReady turns a job description into a printable interview practice handout for placement batches. Each question comes with "what a strong answer covers" points and a follow-up question, so students practise explaining their thinking rather than memorising answers. A trainer can add their own questions, and the handout downloads as a Markdown file. The people who would pay are placement cells and training institutes that run prep batches every season.

Demo

Code

Being-RAYtastic / placeready

Generate interview practice handouts for placement batches from a job description, using AI Model running locally through Ollama.

PlaceReady

Generate interview practice handouts for placement batches from a job description, using Gemma 3 4B running locally through Ollama.

Setup

1. Install Ollama and run

ollama pull gemma3:4b

2. Install python dependencies

pip install -r requirements.txt

3. Run the app

streamlit run app.py

NOTE

• config.py: All settings (like MODEL_NAME) contain in this file. So you can change them without touching the rest of the code

View on GitHub

How I Built It

PlaceReady uses Gemma 3 4B, an open-weight model, running locally through Ollama. The Python app is split into separate files: config.py for settings, prompts.py for prompt text, llm.py for the only code that talks to Ollama, and app.py for the Streamlit page. The model only sees the job description and the trainer's questions, and it runs entirely on the trainer's machine. I improved the output by iterating on the prompt, removing example answers by default and requiring concrete points in each "what a strong answer covers" list.

Why Does Open Innovation Matter?

A placement trainer often works with internal hiring material and student details that shouldn't go to a third-party service. With an open-weight model running locally, that text never leaves the trainer's machine, and there's no per-query fee, so a training institute can run it for every batch. Because the prompts and model are open, a trainer can change them to match their own rounds, which a closed API doesn't allow.

Prize Categories

Best Use of Gemma (Gemma 3 4B is the model that generates every question and point)