Drop Nikolina into Claude and get an ML platform engineer who kills training-serving skew at the feature layer instead of letting every team rediscover it in production. Nikolina builds the platform ML teams work on: feature stores including Feast, Tecton and the Databricks Feature Store; offline versus online store design; point-in-time correct joins and the training-serving skew they prevent; feature versioning, reuse and discovery; materialization and freshness SLAs for online features; model serving infrastructure across batch, real-time and streaming inference; autoscaling, latency budgets and GPU scheduling and cost; experiment tracking and the model registry as platform services rather than per-team scripts; multi-tenant platform design with self-service paved paths; and a clear boundary between what the platform owns and what each product team owns. What you get →Offline and online feature stores, point-in-time correct joins →Training-serving skew detection and prevention →Serving infra: batch, real-time, streaming, GPU scheduling →Model registry, experiment tracking and multi-tenant paved paths 📄 nikolina-feature-store-ml-platform-engineer.skill Under 2 min install Works with Claude, ChatGPT & any AI chat How to install Download the .skill package → open Claude → paste SKILL.md into your Project Instructions or system prompt → describe your requirement → Nikolina builds the answer. Includes a full worked example so you see exactly what you get.
