- Reproducible Research
- Unconstrained computational resources
- Low bandwith requirements
- Generic Ansible playbooks that will spin up ec2 servers with required resources and tools
- Data will be stored in s3 or in EBS data volumes
- Ipython notebooks for interactive exploration and analysis (jupyter?)
- Ipython notebooks for api creation to expose data to other sources (or flask perhaps)
- playbook to set up a production predictionio env
- jupyter api post