stefanwebb/foundation-models-time-series

This project is a collection of educational materials for a specific webinar (ODSC Jan 2025). With only 3 stars and a derivative nature, it lacks any structural or technical moat. It serves as a pedagogical wrapper around existing foundation models like Amazon's Chronos, Google's TimesFM, or Salesforce's MOIRAI. The defensibility is minimal because the value lies in the explanation of other people's work rather than original IP. From a competitive standpoint, frontier labs like Amazon and Google are the primary threats as they not only build the models featured here but also provide their own managed services (e.g., AWS Forecast) and comprehensive documentation that renders third-party tutorials obsolete quickly. The displacement horizon is short because time-series foundation models are evolving rapidly; a tutorial from early 2025 will likely be outdated by mid-2025 as newer architectures or more efficient fine-tuning techniques emerge.