☀️ Summer of Agents 2026 applications — three graduate-student research slots pairing you with coding agents on real Aegean AI problems. Read the overview →
Milestones assume a project kick-off date of June 3, 2024. In addition the plan below accounts for the relatively sequential nature of this project. For example, work on model training milestones cannot be done in parallel with the data pipeline milestone. In addition, the dates below are estimated and it is understood that future milestones can start earlier or later depending on the progress made.
Data pipeline where training, validation and test datasets are stored in object store or other persistent storage facility.Data pipeline source code provided in the private GitHub repository.
Multiple models developed and performance validated are stored in the model store.Model training source code provided in the private GitHub repository.
Serving pipeline runs on a hyper-edge compute environment and event streams are produced. Does not include the consumption of the events by the edge computer.Serving pipeline source code is provided in the private GitHub repository.
Inference event streams are consumed and seam quality alarm is triggered. Alarm trigger is consumable by other systems or applications to enable visual or audio alerts to machine operator(s).
