Building a serverless AI powered translationservice
summary
A working Slack bot gets text, auto-detects language with Comprehend and calls Amazon Translate (70+ languages) inside a Step Functions serverless pipeline using Map states and SDK integrations so almost no Lambda code is needed. Translated text lands in S3, an EventBridge event triggers a Step Function that starts a Amazon Polly speech synthesis task (neural engine) producing MP3s. The live demo runs end to end with translations and voice files generated and returned to Slack.
AI-written summary · about
chapters (20)
- 0:00Serverless AI translation intro
- 2:07Introduction
- 3:13Generative vs traditional AI
- 6:48AWS AI services overview
- 7:52What we're building
- 8:13Architecture and services
- 12:10Building translation service
- 19:41Storing text in S3
- 21:57EventBridge event
- 23:03Testing translation service
- 24:36Blog translation use
- 25:22Text-to-speech service
- 26:05Voice pipeline config
- 29:56Waiting for synthesis
- 30:40Handling completion
- 32:09Slack voice delivery
- 33:13Testing voice generation
- 35:48Demo results and limits
- 36:12Blog real-world app
- 36:53Summary
related talks
48mBuilding Serverless AI powered Applications on AWS
44mBuilding a serverless event driven Slack Bot
30mThinking Serverless: From User Request to Serverless Solution
19mA Serverless State of Mind
56mBuilding a Serverless, EventSourced Slack clone
39mThinking Asynchronously
22mS3 & Lambda - flexible pattern at the core of serverless applications
50mBuild, Deploy, Debug and Tune Serverless Microservices