Products
Team projects
Allganize AI Messenger SDK
At Allganize, I developed the AI messenger SDK provided to clients and internal tools for managing model training.
3SDK layers: core logic, feature API, and interface
- Period
- 2022.09 ~ 2023.03
- Organization
- Allganize Korea
- Role
- SDK layer design and training management features
- Key features
- Clients could build their own designs and features on the shared SDK, while model training history could be managed by version.
- Work and results
- Model training history versioned with support for restoring earlier versions
- SDK separated into core logic, feature API, and interface layers
- Real-time training progress notifications
- Tech stack
- GraphQL
- iframe
Context
At Allganize, I developed internal model training tools and the AI Messenger SDK provided to clients. The internal tools kept a history of model changes and training, with real-time progress updates.
Clients needed different designs and features in the messenger. I separated core logic, APIs, and the interface so they could extend the parts they needed while keeping shared functionality.
Decisions and implementation
- AI model training system: I built version management for each model modification, training, and generation process.
- I used GraphQL subscriptions to send real-time notifications of training progress.
- AI Messenger SDK: I designed a modular three-layer architecture (Core, Headless API, and UI) so each feature could be extended independently.
- I built a CSS isolation environment on top of iframe isolation.