Project overview
Librifly.com is a content platform selling downloadable images, art, stories, poetry, maps, posters, notebooks and books. We have built a complete AI automation system — all Python code is written and ready — and need an experienced developer to deploy, configure, and test it on our existing Hetzner Linux server running Nginx and MySQL.
This is a deployment and configuration job, not a build-from-scratch project. The codebase is clean, documented, and structured across four deliverable packages (Sprints 1–4). Your job is to get it running correctly on our server and verify that the full pipeline works end to end.
What the system does
The automation pipeline:
1. Scrapes content from librifly.com (images, art, stories, poetry, maps)
2. Clusters content into niche topics (e.g. Monet, Vintage France, New York Art)
3. Uses the Claude AI API to write original editorial articles in 9 languages
4. Generates complete static HTML sub-sites served on wildcard subdomains (e.g. monet.librifly.com)
5. Populates each sub-site with real product listings and pricing from librifly.com
6. Publishes automated posts to Instagram, Facebook, X and Pinterest when each sub-site goes live
7. Rolls out one new language version per day (English → Spanish → French → German → Italian → Portuguese → Chinese → Japanese → Korean)
Server environment
- Hetzner Cloud server, Finland
- Linux (Ubuntu)
- Nginx (already running)
- MySQL (already running, existing database)
- Python 3.12 (to be installed)
- Redis (to be installed — lightweight, standard apt install)
- Static IP, domain already configured
Technical scope
Sprint 1 — Core pipeline
- Install Python 3.12, Redis, virtualenv, all pip dependencies
- Configure `.env` file with API keys and DB credentials
- Run 4 SQL migrations to create required tables
- Configure Nginx wildcard subdomain (`*.librifly.com → server`)
- Obtain wildcard SSL certificate via Let's Encrypt (DNS challenge)
- Set up systemd services for Celery worker and beat scheduler
- Test end-to-end: scrape → discover niche → generate HTML → serve on subdomain
Sprint 2 — Shop layer
- Run SQL migration
- Test product scraper against live librifly.com product pages
- Adjust CSS selectors if needed (product page HTML inspection)
- Verify product pricing and buy-links render correctly in sub-site
Sprint 3 — Social media agents
- Create Meta developer app (Instagram + Facebook Graph API)
- Obtain and configure long-lived Meta access token
- Set up X (Twitter) developer account (pay-per-use plan)
- Configure Pinterest developer app and OAuth2 token
- Set up Hetzner Object Storage bucket (S3-compatible, for Instagram image hosting)
- Test a post on each of the four platforms
Sprint 4 — Multilingual rollout
- Run SQL migration
- Test language builder for one language (e.g. Spanish)
- Verify hreflang tags in HTML output
- Verify CJK (Chinese, Japanese, Korean) font loading
Required skills
- Python 3.10+ (Celery, Redis, MySQL connector, Requests, BeautifulSoup, Jinja2, Boto3)
- Linux server administration (Ubuntu, systemd, file permissions)
- Nginx configuration (virtual hosts, wildcard subdomains, SSL)
- Let's Encrypt / Certbot (wildcard certificates via DNS challenge)
- MySQL (running migrations, basic administration)
- REST API integration (Meta Graph API, X API v2, Pinterest API v5)
- Web scraping (BeautifulSoup, CSS selector debugging)
- Object storage / S3-compatible (Hetzner or AWS experience)
- Git (project is delivered as zip files; basic git workflow preferred)
Nice to have
- Prior experience with Meta Graph API app setup and token management
- Celery + Redis in production
- Hetzner Cloud or Hetzner Object Storage experience
- Experience with multilingual / hreflang SEO setups
Deliverables
1. All four sprints fully operational on the Hetzner server
2. At least one complete sub-site visible at a `*.librifly.com` subdomain
3. Celery worker and beat scheduler running as systemd services (surviving reboots)
4. At least one test post successfully published on each social platform
5. Brief handover notes covering: token expiry dates, how to add new niches, how to trigger a manual pipeline run
Estimated effort
Realistic: 25–30 hours over 3–4 working days.
The largest single time block is Meta developer app setup (3–5 hours) due to Meta's developer portal complexity. All other sprints are 1–3 hours each.
Budget
Fixed-price quotes are welcome if you are confident in the scope.
How to apply
Please include:
- Brief description of similar deployments you have done (Python automation, social API integrations, Linux server setup)
- Your experience with Meta Graph API specifically — this is the most complex part of the project
- Your availability to start and realistic completion timeline
- fixed-price quote
We will share the full codebase (four zip files) with shortlisted candidates before final selection so you can review the code before committing.
... Show more