Assembly AI provides Industry leading Speech AI models that are focused in understanding human speech and converting that to text with the highest accuracy using machine learning. Thousands of organizations leverage Assembly AI's models to build cutting edge AI products and features that serve millions of end users around the world. In this architecture discussion, they dive deep into how they have built a complete AI/ML platform to take voice data (such as calls, virtual meetings, and podcasts) and transcode them to text. In this pipeline, they can do speaker detection, sentiment analysis, chapter detection, PII redaction, and more. With this architecture, they address three challenges, 1/ Scalability – handle millions of requests per day, 2/ Ease of ML model deployment and iteration based on job type and 3/ Security and compliance. The implementation includes AWS services like Amazon ECS for orchestration, Amazon SNS for pub-sub, Amazon SQS for job queuing and Amazon ECS for model deployment. Learn more about AssemblyAI by watching our Product Overview on YouTube - https://www.youtube.com/watch?v=UT1sBCuSJxE Read about AssemblyAI's punctuation restoration and truecasing models to support scalability - https://www.assemblyai.com/blog/introducing-our-new-punctuation-restoration-and-truecasing-models/ Learn more about AssemblyAI's enterprise-grade security practices to keep your data safe - https://www.assemblyai.com/security Check out more resources for architecting in the #AWS​​​ cloud: http://amzn.to/2ZIbygO #AWS #AmazonWebServices #CloudComputing #ThisIsMyArchitecture #AmazonTranscribe