Telegram AI Video Bot

I need a Python-based agent that lives inside Telegram and guides a user through this flow:
1. Prompt the user to drop a product photo.
2. Read the image, pass the concept to Synthesia for a professional-looking avatar voice-over, then stitch the result with a Runway Gen-4 background so the final clip runs no longer than 30 seconds.

My Synthesia and Runway keys are ready; just wire them through an .env file and keep all sensitive data outside the repo. Please structure the project with a clear /src folder, separate config module, requirements.txt, and a short Makefile or shell script that spins everything up inside a virtual environment.

Once the video is rendered and approved by the user the bot should:
• Automatically push it to Instagram, Facebook and Twitter via their latest APIs.
• Retrieve post analytics—views & likes, comments & shares, plus follower growth—and log them to a lightweight SQLite or CSV store.
• Offer a simple rule or cron job to replace the video or add fresh variants when engagement drops below a threshold I can edit in config.

Suggestions on scheduling posts, hashtag strategy, and rate-limit handling are welcome; please document them in README.md.

Deliverables
• Clean, well-commented Python code (PEP 8) organised as explained above
• Telegram bot token setup instructions
• Sample .env.template and config.yaml
• README with steps for running locally, deploying on a small VPS, and hooking up to each social network
• A short note on how to extend the metrics tracker to other KPIs later

I’ll test by uploading a sample image, checking that the bot returns a 30-second professional video, seeing it published on all three channels, and confirming that the metrics file updates after an hour.

I would like to have these in quick turnaround time and looking a long-term reliable partners.