Contributor to Daggit, an open-sourced end-to-end machine learning development platform to help iterate, train models, deploy, and schedule ML pipelines.
Daggit was conceptualized as part of Project Sunbird to streamline the work of the Machine Learning team and allow Data Scientist, ML Engineers to quickly test and productionise different models. The project has been architectured to be generic (not specific to current Sunbird use-cases) and open-sourced along with multiple data pipelines and ML service.
Daggit features:
- Fast error-proof productionisation
- Handle Scalability vs. flexibility trade-off
- Reusability & extensibility of modeling effort
- Management of offline modeling experiments
- Interactive monitoring of modeling experiments
- Search and Discover Apps
- Assured Product quality
- Schedule, Monitor and Manage jobs
- Interoperable ML
Different components for ML development on daggit.