I'm a 17-year-old IT student from Luxembourg. A few months ago I got tired of spending 2-3 hours a day manually browsing Upwork, Malt, and Freelancer looking for projects.
So I built an automation system that does it for me — 24/7, on a Raspberry Pi 3.
Here's what it does, how I built it, and every painful lesson I learned along the way.
What the system does
- Scans Upwork, Malt, and Freelancer every 30 minutes
- Scores each job 0–100 with AI based on my profile
- Generates proposals in English, French, and German
- Sends the best jobs to Telegram with inline A/B buttons
- Tracks which proposal style gets more replies
- Sends daily stats and weekly market trend reports
- Reminds me to follow up after 3 days
The stack
n8n — self-hosted workflow automation (Docker on Raspberry Pi 3)
Groq API (Llama 3.1-8b-instant) — AI scoring and proposal generation
Supabase — PostgreSQL database for jobs, proposals, clients
SerpAPI — searching job boards via Google
Apify — scraping Upwork listings
Telegram Bot API — alerts and bot commands
Cloudflare Tunnel — HTTPS for webhooks
Total running cost: ~$5/month.
7 workflows
01 - Job Discovery — runs every 30 minutes, searches 10+ sources, deduplicates via Supabase unique constraint on URL
02 - Proposal Generator — AI scores the job, generates two proposal variants (formal vs hook-first), sends to Telegram with A/B buttons
03 - Follow-up Reminders — checks Supabase every 3 days for unanswered proposals
04 - CRM via Telegram — full client management through bot commands (/jobs, /stats, /clients)
05 - Market Intelligence — daily report: how many jobs found, average score, top platforms
06 - Trend Analysis — weekly report on what skills are trending in automation
07 - Lead Generation — finds companies actively using Zapier or Make who might want to switch to n8n
Lessons learned (the hard way)
1. Cyrillic text breaks JSON body nodes silently
If you have Cyrillic characters in a JSON body field with newlines, n8n throws a "Bad control character" error. Keep everything on a single line or use \n instead of real line breaks.
2. Multiple inputs into a Code node need a Merge node first
When two HTTP Request nodes feed into one Code node, use a Merge node (Combine by Position) in between. Otherwise $('NodeName') throws "hasn't been executed" errors.
3. Cloudflare tunnel URL changes on every restart
This breaks all Telegram webhooks. I solved it with a systemd service that restarts cloudflared and updates the WEBHOOK_URL environment variable in docker-compose automatically.
4. Deduplication belongs in the database, not in n8n
Adding a UNIQUE constraint on jobs.url in Supabase and setting the insert node to "Continue on Fail" is cleaner than trying to check duplicates inside the workflow.
5. Code nodes must always return items
If a Code node returns nothing (empty array or undefined), the workflow silently passes dummy data downstream. Always return [] or a proper item.
6. LLMs ignore "return only JSON" instructions
Groq kept wrapping JSON in
json markdown blocks. Fix: strip the backticks before parsing, or use separator-based output instead of JSON.
---
## A/B testing proposals
This was the most interesting part to build. Instead of one proposal, the system now generates two variants with different tones:
- **Variant A** — formal, structured, under 100 words
- **Variant B** — hook-first, starts with the client's problem, under 70 words
Telegram sends both with buttons. When I tap which one I sent, the result gets stored in Supabase. Over time I'll have data on which style works better for which type of job.
---
## GitHub
All 7 workflow files (sanitized — API keys replaced with placeholders) are on GitHub:
👉 https://github.com/ihorbezruchko/n8n-freelance-autopilot
If you want the full working setup with the complete configuration guide:
👉 https://n8n-freelance-autopilot.netlify.app
---
Happy to answer questions about any part of the implementation. What would you build differently?
SOCIAL SHARE CARD GENERATOR