Building a Machine Learning Enabled Bot in the Cloud

on YouTube ↗01h 21m

" summary

A working serverless Slack bot that classifies aircraft images using AWS Lambda, API Gateway, S3 and a SageMaker endpoint plus optional Amazon Rekognition is built and demoed. The quick demo trains on ~1,500 images (≈475 per class) for 5 epochs and shows validation around 85–92% in ~7–8 minutes; production used ~10,000 images per class on a GPU cluster and ran for hours. Integration uses Lambda + API Gateway, it generally works but accuracy and cost depend on training scale and running endpoints.

AI-written summary · about

" chapters (9)

1. 0:00Introduction
2. 3:57Aircraft ML Twitter bot
3. 7:11AWS architecture
4. 10:01Setting up Amazon SageMaker for model training
5. 35:22Lambda interface
6. 54:53API Gateway interface
7. 1:00:40Slack bot integration
8. 1:14:34End-to-end testing
9. 1:17:31Cleanup and Q&A

" related talks

• 41mBuilding Serverless Machine Learning Models in the CloudAlex Casalboni · PyData DC 2016
• 45mServerless Machine Learning: a beginner's guideBret McGowen, Yufeng Guo · Devoxx Belgium 2017
• 55mMachine Learning and AI in Java on AWSJulien Simon · JavaDay Ukraine 2017
• 48mBuilding Serverless AI powered Applications on AWSAdrian Hornsby · Devoxx Belgium 2017
• 44mPragmatic Machine Learning in the CloudJoshua Arvin Lat · Conf42 Machine Learning 2021
• 01h 41mGetting started with Machine Learning on AWS.Julien Simon · Devoxx Ukraine 2020
• 34mBuilding Machine Learning-powered Applications with JSJoshua Arvin Lat · Conf42 JavaScript 2022
• 26mServerless Architecture for Product Defect DetectionMohsin Khan · Conf42 Machine Learning 2022