Introduction to NEXUS

Build, evaluate, and operate ML models on the NEXUS platform — from first fit to production lifecycle.

9 lessons 3 phases beginner

Get started Signed in as your account.

what you'll build

what you walk away with.

01 A trained NEXUS credit-risk classifier with a persistent server-side model ID.
02 Probability-based predictions with your own threshold choice, and model reload via load_model().
03 An enriched model trained on multi-table joins and categorical features.
04 A data-driven TOP_FEATURES selection from cumulative feature importance.
05 An async submit/poll training workflow you implemented yourself.
06 Retry-with-backoff resilience around NEXUS's typed exception hierarchy.
07 A production model-lifecycle toolkit: attributes, inventory, deletion, per-env keys.

curriculum

9 lessons across 3 phases.

phase A · Fundamentals 3 lessons

01 What Is Prediction?
02 Your First Credit Risk Model
03 Probabilities and Persistence

phase B · Features and production patterns 3 lessons

04 Smarter Features
05 Feature Importance
06 Async Patterns

phase C · Operations 3 lessons

07 Resilient Pipelines
08 Diagnostics and Debugging
09 Operating at Scale

prerequisites

what you need before you start.

Python + pandas: Comfortable loading CSVs into DataFrames and basic column operations. Lesson 1 is a gentle ML on-ramp; later phases assume train/holdout splits and basic classification metrics.
NEXUS API key: Provided for the workshop (format ak_...) — models you train live server-side under it.
Cloudsmith token: Your token for installing the private fundamental-client SDK.
No toolchain setup: Nothing to install by hand — setup builds the Python environment for you, interpreter and package tooling included.

start

how to start.

Once Claude Code is set up, tell it “set up the nexus workshop” and it clones the project and queues the first lesson.

New here? The getting-started guide walks you through installing the lwc CLI and the Claude Code plugin once — everything you need before that line works.