Public software, model records, and technical writing, with active research clearly separated from released work.
SagitariusAI keeps a clear boundary between public work and research that is still developing. Evidence appears in repositories, model records, and technical publications. Professional credentials provide supporting context.
GitHub contains our public software, documentation, and release records. It is the primary location for examining how released software is structured and maintained.
Hugging Face contains public model work and supporting documentation. When we release a model, its record should explain the intended use, evaluation, and known limits.
The SagitariusAI portfolio brings together selected work in software, models, and evaluation. It also records relevant infrastructure and communication work. Its purpose is to provide a navigable view of demonstrated capability.
Current research includes model curricula, agent coordination, and local execution. Other work concerns datasets, evaluation, and AI-assisted software delivery. Research may be described publicly without being presented as a released product or commercial commitment.
Technical articles and post-mortems document the method, the result, and failures when they occur. They also record revisions and remaining uncertainty. This provides context that a finished artifact alone cannot supply.