A user-friendly ML platform that lets non-experts upload datasets, receive model suggestions, schedule training, and deploy models for real-time predictions
0
stars
0
forks
source link
Deployment - Training server for training models #47
As a user, I want to be able to queue and train new models that I can use later for inference.
Persona:
Ken, the startup founder, wants to train a new model on a dataset he uploaded.
Feature:
A training server that polls from the SQS queue and trains models one by one
Business Value:
Users will (finally) be able to queue model training jobs.
Tasks
Set up a server that connects to the SQS queue
Poll the SQS queue and train models one by one
Upload the models to the s3 bucket and add a record to the relational database
Deploy the server to the cloud
Acceptance Criteria
The server will only be able to train models one by one
Acceptance Tests
A server that polls the SQS queue and trains models one by one. You should be able to schedule a new training job, and after some time a new model should be trained and uploaded.
User Story:
As a user, I want to be able to queue and train new models that I can use later for inference.
Persona:
Ken, the startup founder, wants to train a new model on a dataset he uploaded.
Feature:
Business Value:
Users will (finally) be able to queue model training jobs.
Tasks
Acceptance Criteria
The server will only be able to train models one by one
Acceptance Tests
A server that polls the SQS queue and trains models one by one. You should be able to schedule a new training job, and after some time a new model should be trained and uploaded.