Skip to content
Back to the record

Job Posting Validity Checker

A web app that checks a job posting against Ontario's 2026 job posting transparency rules and returns a valid or invalid verdict, with the reasons.

Seneca Software Engineering Competition · Dec 2025 · 1st place

The user flow: pick the form or text box tab, fill it in, check, then fix and recheck until the posting is valid

Brief

Does this posting follow the new Ontario rules?

Ontario's 2026 transparency rules ask job postings for a pay range, the employment type, and a disclosure when AI is used to screen applicants, and they bar Canadian-experience requirements. The tool takes a posting as a structured form or as pasted text, and returns VALID or INVALID with a transparency score, a confidence level, and the fields that failed.

My part

The frontend flows, and a stack that starts with one command.

I built the two submission flows and wired them to the backend: a structured form and a raw-text box, each posting to its own FastAPI endpoint and feeding the same result panel.

I containerized the backend and frontend with Docker and wrote the Compose file, so the whole app starts with docker compose up --build.

How it decides

Hard rules first, a model second.

A rules engine runs first. Missing required fields, an out-of-range salary, an undisclosed use of AI, or a Canadian-experience clause makes a posting invalid on the spot. If nothing fails, a logistic regression text classifier and a random forest transparency scorer, both trained with scikit-learn, rate the posting.

Minsik Kim built the backend and both models. Khai Ngo designed the interface.

The wireframe and palette: the job posting form, a Check button, and the result panel