Advanced Spark and Tensorflow Meetup - London - Nov 15, 2016 - Deploy Spark ML and Tensorflow AI Models from Dev Notebooks to Prod Microservices
4 likes743 views
The document outlines a meetup held on November 15, 2016, focusing on deploying Spark and TensorFlow models into production, highlighting the impact of Pipeline.io and updates on Spark and TensorFlow. It shares metrics on the growing community, successful workshops, and future plans for performance and expansion in 2017. The session included demonstrations on model deployment, scaling, and dynamic code generation.
Advanced Spark and Tensorflow Meetup - London - Nov 15, 2016 - Deploy Spark ML and Tensorflow AI Models from Dev Notebooks to Prod Microservices
- 1. ADVANCED SPARK AND TENSORFLOW MEETUP Deploy Spark andTensorflow Models: From Notebook Dev to Microservice Prod London - November 15, 2016 ThankYou, PipelineIO,Rise, Barclays! http://pipeline.io
- 2. MEETUP AGENDA âą Meetup Updates, Metrics, and Announcements âą Spark Updates âą Tensorflow Updates âą Deploying and Scaling ML Models from Notebook to Production with PipelineIO (Chris Fregly, PipelineIO)
- 3. WHO AM I? Chris Fregly -------- Research Scientist @ PipelineIO (Formerly Netflix and Databricks) -------- http://pipeline.io
- 4. WHO ARE YOU? --Techies of London --
- 5. FUN WORKSHOP LAST MONTH (HERE IN LONDON)
- 6. MEETUP METRICS âą 10,000+ Members Globally in 1Year! âą 5,000 Members in San Francisco (Home) âą Github Repo: 900 Stars, 300 Forks âą DockerHub Repo: 6,200 Pulls
- 8. SPARK SUMMIT EU 2016 (BRUSSELS) âą https://spark-summit.org/eu-2016/schedule/ âą Structured Streaming (2 talks) âą ML + Structured Streaming (1 talks) âą ML Model Deployment (2 talks) âą Spark Performance (4 talks)
- 10. TENSORFLOW WORKSHOP âą Great Workshop From Google Directly âą Finally, training involving more than MNIST! âą Slides âą http://bit.ly/tf-workshop-slides âą Workshop âą https://github.com/amygdala/tensorflow-workshop
- 11. WHAT IS PIPELINE.IO? ExtendingYour ML Pipelines into Production 100% Open Source! http://pipeline.io
- 12. BRAINSTORMING AND VALIDATING âą Major Gaming Company âą Large Ride Sharing Service âą Popular Q & A Site âą Online Clothing Retailer âą DominantVideo Streaming
- 13. PIPELINE.IO FOCUS âą Model Deploying andTesting âą Model Scaling and Serving âą Online ModelTraining âą Dynamic Model Optimizing
- 14. MODEL DEPLOYING AND TESTING Continuously Test and Deploy Models in Production!
- 15. MODEL SCALING AND SERVING
- 16. ONLINE MODEL TRAINING âą Continuous, Incremental, and Partial Training âą Kafka + Spark Streaming + Spark ML âą Real-time, Dynamic Recommendations
- 17. DYNAMIC MODEL OPTIMIZING Generate Optimized Code from Spark ML!
- 19. PIPELINE.IO PLAN FOR 2017 âą Performance âą Code Generation: CPU and GPU âą Continued Global Expansion
- 20. MORE WORKSHOPS IN 2017
- 21. WEâRE HIRING!! âą Kafka, Spark ML, and TensorFlow Contributors âą Systems Engineers âą GPU/CUDA Engineers âą C++, Java, Scala, Python WE ONLY HIRE NICE PEOPLE!!
- 22. DEMO! Circuit Breakers and Request Batching
- 23. DEMO! Deploy Spark ML DecisionTree to Production Deploy to Cloud or On-Premise!
- 24. DEMO! Dynamic Code Generation of DecisionTree