I am designing an end-to-end system that can take a predefined topic—Travel is the first one on the list—and automatically produce short social-media-ready videos featuring realistic human avatars. The workflow I have in mind is simple for the user but smart under the hood: the engine should look at current trends and audience-engagement metrics before it writes a script, selects visuals, animates the avatar, adds voice-over, and exports the final clip. Preferred stack is flexible as long as it can integrate reliable AI tools for script generation, text-to-speech, avatar animation, and video compositing. Solutions such as Python back-end with OpenAI or similar LLMs, plus any of the established avatar/video APIs (e.g., Synthesia, D-ID, HeyGen) are perfectly acceptable, but I am open to alternatives if they achieve the same quality with realistic human avatars. Deliverables • A working prototype accessible through a simple web dashboard or API • Ability to load a topic list (starting with Travel) and toggle the two criteria—Current trends and Audience engagement metrics—before generation • Auto-generated vertical and square video outputs in HD, ready for TikTok, Reels, and Shorts • Brief documentation covering setup, dependencies, and how to extend the topic library Acceptance criteria 1. The prototype generates at least one 30-60 sec Travel video with a realistic avatar and publishes the file without manual editing. 2. Script, visuals, and call-to-action reflect current trend data pulled at runtime. 3. A metrics log shows which engagement indicators influenced each creative decision. If you have built similar AI-driven content systems or avatar video pipelines, I would love to see a quick demo or code sample to speed up the conversation.