AWS Community Day Greece
The Greek community day at Technopolis in Athens, the old gasworks in Gazi.
On the programme
From Raw Footage to Publish-Ready: Building a Serverless AI Video Pipeline on AWS
2026 · Athens
493 raw clips and 40 to 60 hours of manual editing, and the pipeline I built to stop doing that by hand. S3, Step Functions across six state machines, Rekognition, Transcribe, Bedrock and Bedrock Data Automation, AgentCore and MediaConvert, with eight specialist agents that gate each other.
The services doing the work
Read off the slides, not guessed. Each card says what that service did in this build, and links to the page explaining the capability and how to run it somewhere else.
Amazon S3
Holds the 493 raw clips from phone, Osmo and friends’ drops, and every file derived from them.
AWS Step Functions
Six state machines carry a clip from raw to publish-ready, with idempotency and retry logic.
Amazon Rekognition
Faces, scenes and emotion: what is actually in the picture.
Amazon Transcribe
Word-level transcription with speakers and custom vocabulary.
Amazon Bedrock Data Automation
Scenes, topics and sentiment across the footage.
Amazon Bedrock
Claude and Nova do the deciding: clip scoring, narrative structure and template selection.
Amazon Bedrock AgentCore
Hosts the eight-agent film crew: director, editor, cine, sound, fact, story, pacing and quality.
CrewAI
The framework the eight agents run under, each with one job and the power to gate the others.
AWS Elemental MediaConvert
The final compilation, rendered to 16:9, 9:16 and 1:1 from one VariantSpec.
AWS Lambda
Triggers MediaConvert and carries the small steps between the state machines.
Slides from the talk
From Raw Footage to Published Video: the pipeline that cuts my own clips. These are the slides that carry the architecture.
The whole pipeline in one column: S3 for direct upload, Step Functions for orchestration, Rekognition, Transcribe and Bedrock BDA for analysis, Bedrock with Claude and Nova for decisions, an eight agent film crew on AgentCore, and MediaConvert producing a 16 by 9 YouTube cut, a 9 by 16 reel and a 1 by 1 LinkedIn version. Slide from the VideoFlow deck.
Amazon Bedrock and BDA: why the service is in the pipeline, how it is used on a downsized clip, and the quotas and gotchas, over a data centre scene. Slide from the VideoFlow deck, background image AI generated.
Bedrock AgentCore and AWS Elemental MediaConvert: the multi agent setup with feedback loops and memory, and MediaConvert building the compilations, with the quotas noted. Slide from the VideoFlow deck, background image AI generated.
Raw files in, three versions out: one 90 second 4K clip with no edits going in on the left, three finished formats coming out on the right. Slide from the VideoFlow deck, background image AI generated.
The crew: VideoFlow as director on Step Functions, and the eight specialist agents around it, director, editor, cinematographer, sound designer, fact checker, story analyst, pacing analyst and quality checker. Slide from the VideoFlow deck.
The five takeaways the talk closed on, starting with focusing on a specific task and ending with not expecting AI to do everything. Slide from the VideoFlow deck.
The longer write-ups
Where the builds behind these sessions are explained properly.
Published Video in Under 3 Minutes: How I Built an AI Film Crew on AWS
The longer write-up: the build, the services and what it costs to run.
From Transcript to Finished Deck in One Working Session
The longer write-up: the build, the services and what it costs to run.
What is an AWS Community Day?
The longer write-up: the build, the services and what it costs to run.
10 Years of AWS User Group Vienna
The longer write-up: the build, the services and what it costs to run.
The same talk, other stages
Same session, a different room. Each page carries its own deck.
Where all of this comes from
Every claim on this page traces back to the deck, the recap or the event’s own site.
- Slides: From Raw Footage to Publish-Ready: Building a Serverless AI Video Pipeline on AWS, 2026
- Published Video in Under 3 Minutes: How I Built an AI Film Crew on AWS
- From Transcript to Finished Deck in One Working Session
- What is an AWS Community Day?
- 10 Years of AWS User Group Vienna
Everything here is free to read, use and share. No email needed for the slides. If you reuse something, please credit the source.
This page mirrors the talk page on lindamohamed.com, where it was first published.