Skip to main content
ToolDix logoToolDixFind, compare, and use tools
Sign In
Browse 20 learning paths

AI learning path

MLOps & Model Deployment

Move models from notebooks into reproducible services with versioned data, automated tests, deployment controls, and production monitoring.

For ML engineers, platform teams, and production-minded developers

8 original lessons on ToolDix, roughly 142 minutes end to end

Start the course

Course outline

  1. 1Package a Reproducible ModelTurn a notebook result into an immutable artifact by pinning the inputs teams usually forget, defining the inference contract that lives outside the model file, and drawing a release boundary that promotion cannot cross by accident.Intermediate20 min
  2. 2Monitor Quality, Drift, Latency, and CostOrder monitoring signals by how long each takes to tell you something, choose a drift reference window that can actually fire, and attach a response to every page so the team keeps trusting the alerts.Advanced22 min
  3. 3Experiment Tracking and Model LineagePin the five inputs that make a run reproducible, record runs instead of scores, and build lineage that answers the questions you get during an incident.Intermediate16 min
  4. 4Training-Serving SkewSeparate the four origins of skew, share one feature transform between training and serving, and enforce point-in-time correctness so offline metrics stop lying.Advanced17 min
  5. 5Model Serving ArchitecturesChoose between batch, online, and streaming by acceptable staleness, budget latency across the whole request, and pick the scaling lever that fits the constraint.Intermediate16 min
  6. 6Safe Deployment Strategies for ModelsUse shadow, canary, and blue-green for the different questions each one answers, wire shadow traffic so it cannot affect users, and build the four preconditions for a real rollback.Intermediate16 min
  7. 7Continuous Delivery for Machine LearningBuild one pipeline that handles code, data, and config triggers, layer the ML test pyramid beneath model metrics, and gate promotion on reproducibility, slices, cost, and ownership.Advanced17 min
  8. 8Serving LLMs EfficientlyUnderstand why prefill and decode are different bottlenecks, how the KV cache limits concurrency, and why continuous batching changes the economics of LLM serving.Advanced18 min

Optional, after the lessons

Further reading from primary sources

Published by other organizations and kept here for reference. The lessons above are the ToolDix course; these are where to go once you want the vendor's own documentation.

5 resources

MLOps Course by Made With ML
CourseIntermediate

MLOps Course

Made With ML · Goku Mohandas

An end-to-end course covering product design, data, modeling, testing, reproducibility, serving, observability, and CI/CD for machine-learning systems.

Time
End-to-end course
Author
Goku Mohandas
testingservingobservability
Read the ToolDix guide
MLOps Zoomcamp by DataTalks.Club
CourseIntermediate

MLOps Zoomcamp

DataTalks.Club

A project-led curriculum covering experiment tracking, orchestration, deployment, monitoring, testing, infrastructure, and a production capstone.

Time
9 modules and project
MLflowdeploymentmonitoring
Read the ToolDix guide
Full Stack Deep Learning by Full Stack Deep Learning
Video seriesAdvanced

Full Stack Deep Learning

Full Stack Deep Learning

A systems-oriented course on infrastructure, experiment management, troubleshooting, testing, deployment, monitoring, data management, and team practices.

Time
Lectures and labs
ML systemsdeploymentoperations
Read the ToolDix guide
Rules of Machine Learning by Google for Developers
Classic readingIntermediate

Rules of Machine Learning

Google for Developers · Martin Zinkevich

A field guide to production ML that prioritizes solid pipelines, simple baselines, meaningful features, measurable objectives, and staged system evolution.

Time
43 production rules
Author
Martin Zinkevich
production MLpipelinesmetrics
Read the ToolDix guide
MLflow Getting Started by MLflow
Hands-on labBeginner

MLflow Getting Started

MLflow

Official quickstarts for experiment tracking, model packaging, evaluation, registry workflows, and serving.

Time
Quickstart tutorials
experiment trackingmodel registryserving
Read the ToolDix guide