Upwork Job Alerts with GPT-4.1-mini + Telegram (n8n)

Get Upwork job leads in Telegram—automatically, every 15 minutes

This n8n workflow searches Upwork for new job postings on a 15-minute schedule, uses GPT-4.1-mini to score fit and draft a proposal opening, and instantly sends high-scoring opportunities to your Telegram—while preventing duplicate alerts using an n8n Data Table.

What this workflow does

• Runs on a schedule every 15 minutes to fetch the latest jobs.
• Loads dynamic search settings by downloading a profile.md file over HTTP, extracting the required JSON configuration block, and applying defaults/validation.
• Queries Upwork via GraphQL: builds one Upwork GraphQL search request per configured “search lane” using Upwork OAuth2.
• Normalizes and filters results using deterministic rules (job age, budget floors, blocked keywords), deduplicates within the run, and skips anything already stored in upwork_seen_jobs.
• Uses an AI scoring step with OpenAI (chat model) to rank remaining jobs and draft a proposal-opening candidate.
• Applies score thresholds: only jobs above the threshold receive a formatted HTML alert in Telegram.
• Prevents reprocessing by upserting every processed job (alerted or not) into the upwork_seen_jobs Data Table.

Use cases

• Freelancers and agencies who want continuous Upwork sourcing without manually checking listings.
• SaaS operators and automation engineers who need a reliable lead-triage pipeline with deduplication.
• Anyone tuning Upwork search lanes and AI thresholds to reduce noise and focus on high-fit jobs.

Technical details (n8n)

• Nodes used: if, set, code, telegram, data table, sticky note.
• Upwork integration: GraphQL API with Upwork OAuth2.
• AI scoring: OpenAI chat model (GPT-4.1-mini) to score fit and draft a proposal opening.
• State management: n8n Data Table upwork_seen_jobs automatically created/reused to avoid duplicate Telegram alerts.
• Setup requirement: host profile.md and provide its raw URL in runtime settings.