This document discusses Randall Hunt's Twitter bot @WhereML, which uses Amazon SageMaker and AWS Lambda to determine the location from photos tweeted at the bot. It was built using the LocationNet model trained on over 33 million geo-tagged images. The architecture uses API Gateway to invoke a Lambda function when tweets are sent to @WhereML. The Lambda function calls a SageMaker inference endpoint running the LocationNet model to classify the image location, then posts the results back to Twitter. Details are provided on the model architecture, infrastructure components, and code snippets from the Lambda function.