Business impact

<aside> 💡

The problem: manually checking six job boards ate hours every day, and most listings still turned out to be a mismatch on stack or level.

What it looked like before: each board was checked separately, vacancies were read one by one, filtering ran on manual keyword matching, and it was hard to tell a relevant listing from an irrelevant one without reading the full description.

What was built: an n8n pipeline that pulls vacancies from six sources at once, scores each one through an LLM that compares the listing against a real skills profile instead of matching keywords, and pushes a digest of high-scoring vacancies to Telegram automatically.

Result: the pipeline runs twice a day unattended, filters out irrelevant listings before time is spent reading them, and also monitors Gmail for recruiter emails.

</aside>

A personal tool for automating vacancy search, built on n8n and LLM-based scoring.

Project details

Context:

• Goal: searching for Junior/Trainee AI Automation vacancies across six job boards at once.
• Process before automation: manually checking DOU, Work.ua, Jooble, RemoteOK, WeWorkRemotely and the n8n Community Forum separately, reading every listing, logging matches by hand.

Problem:

• A lot of time went into reviewing vacancies that were obviously not a fit.
• Keeping one deduplicated list across boards was hard.
• Keyword search missed relevant listings with different wording and surfaced irrelevant ones.
• There was no single place showing both new vacancies and recruiter emails.

Solution

<aside> 🧠

An n8n pipeline pulls vacancies from six sources, cleans and deduplicates the list, scores each vacancy through an LLM (Groq) against a real skills profile, writes everything to Google Sheets, and sends a Telegram digest for high-scoring vacancies.

Tools used: n8n, Groq (Llama 3.3 / GPT-OSS), Google Sheets API, Telegram Bot API, Gmail API.

</aside>

Solution logic: