Skip to content
Jobs Netverks

Track

ai-platform

Build Your Own AI Platform

36 lessons: registry, training jobs, inference, auth, cost, eval gates, and rollback—plus 108 MCQs.

Mode
none
Practice
Read / quiz
Lessons
36 units
Start lesson 1 → A platform is not a model

Before you start

The system around models: tenants, a registry, training jobs, inference, evaluation gates, cost, and rollback.

A weight file on a laptop is not a platform. Production needs an owner, a version, a log, and a way to turn the model off.

Internal ML platforms, multi-tenant products, and teams deciding what to build versus buy.

Specify one model’s path from job to production pointer. Name the tenant boundary and the off switch before you add infrastructure.

After you can train and evaluate one model—when more than one person or customer will depend on it.

Lesson order

Sequential — follow top to bottom

36 lessons are live in this track. Start from step 01 for the smoothest path.

  1. 01 intro A platform is not a model

    beginner

    Open →
  2. 02 users-and-tenants Users and tenants

    beginner

    Open →
  3. 03 model-registry A model registry

    beginner

    Open →
  4. 04 artifact-store Artifact storage

    beginner

    Open →
  5. 05 training-jobs Training jobs

    beginner

    Open →
  6. 06 inference-api An inference API

    beginner

    Open →
  7. 07 batch-inference Batch inference

    intermediate

    Open →
  8. 08 feature-store Features for training and serving

    intermediate

    Open →
  9. 09 data-pipelines Data pipelines

    intermediate

    Open →
  10. 10 labeling Labeling operations

    intermediate

    Open →
  11. 11 experiment-tracking Experiment tracking in the platform

    intermediate

    Open →
  12. 12 evaluation-service An evaluation service

    intermediate

    Open →
  13. 13 promotion Promotion is a release

    intermediate

    Open →
  14. 14 config Config separate from weights

    intermediate

    Open →
  15. 15 gateways A gateway in front of models

    intermediate

    Open →
  16. 16 auth Authentication and authorization

    intermediate

    Open →
  17. 17 quotas Quotas

    intermediate

    Open →
  18. 18 observability Observability

    intermediate

    Open →
  19. 19 logging-prompts Logging prompts and predictions

    advanced

    Open →
  20. 20 cost-controls Cost controls

    advanced

    Open →
  21. 21 gpu-scheduling Scheduling GPUs

    advanced

    Open →
  22. 22 queues Queues and workers

    advanced

    Open →
  23. 23 versioning Versioning policy

    advanced

    Open →
  24. 24 rollback Rollback

    advanced

    Open →
  25. 25 ab-tests A/B tests and shadows

    advanced

    Open →
  26. 26 safety-filters Safety filters

    advanced

    Open →
  27. 27 human-review-queue Human review

    advanced

    Open →
  28. 28 secrets Secrets

    advanced

    Open →
  29. 29 multi-env Dev, staging, production

    advanced

    Open →
  30. 30 slas Service levels

    advanced

    Open →
  31. 31 billing Billing and showback

    advanced

    Open →
  32. 32 admin-console An admin console

    advanced

    Open →
  33. 33 incident-response Incidents

    advanced

    Open →
  34. 34 build-vs-buy Build versus buy

    advanced

    Open →
  35. 35 interview Interview review

    advanced

    Open →
  36. 36 capstone Platform spec

    advanced

    Open →