AI Video Workflow Automation
Budget: €250 – €750 EUR
I’m building an end-to-end pipeline that will find videos on my NAS and upload it to another video platform, the important things are:
• Encode: Transcode every incoming file into the adaptive formats my player expects (H.264/H.265 + multiple bit-rates).
• Legal docs: Locate the matching licence, model release or rights agreement in our internal store (or an external source I’ll specify), attach it as metadata, and flag if anything is missing.
• AI analysis: Before upload, run sentiment analysis and scene segmentation on the video so I can generate chapter markers and compliance notes.
- encode the video with ffmpeg to be between 13 and 20 minutes
• Upload: Push the processed file, metadata and AI insights to my platform via its REST/GraphQL endpoints and confirm success.
Infrastructure preference
I want this stack self-hosted, using a local LLM running in a Windows VM where i can monitor it but it runs fully self hosted. It can call other AI's ofcourse to make the task go better and more efficient
Acceptance criteria
1. One command (or scheduled trigger) moves a new file through the entire workflow without manual touch.
2. Logs show each stage, with clear errors if legal docs are missing or analysis fails.
3. The platform does not have an API so the AI must detect all elements and upload all content to the right place for EG 2257 docs but also the video content.
• Encode: Transcode every incoming file into the adaptive formats my player expects (H.264/H.265 + multiple bit-rates).
• Legal docs: Locate the matching licence, model release or rights agreement in our internal store (or an external source I’ll specify), attach it as metadata, and flag if anything is missing.
• AI analysis: Before upload, run sentiment analysis and scene segmentation on the video so I can generate chapter markers and compliance notes.
- encode the video with ffmpeg to be between 13 and 20 minutes
• Upload: Push the processed file, metadata and AI insights to my platform via its REST/GraphQL endpoints and confirm success.
Infrastructure preference
I want this stack self-hosted, using a local LLM running in a Windows VM where i can monitor it but it runs fully self hosted. It can call other AI's ofcourse to make the task go better and more efficient
Acceptance criteria
1. One command (or scheduled trigger) moves a new file through the entire workflow without manual touch.
2. Logs show each stage, with clear errors if legal docs are missing or analysis fails.
3. The platform does not have an API so the AI must detect all elements and upload all content to the right place for EG 2257 docs but also the video content.