Free Learning Roadmap

How to Become a ML Engineer

Deploys, scales, and monitors ML models in production. The bridge between research notebooks and real user impact.

Topics
12
Resources
36
Cost
Free
Most resources are free or freemium
Format
Self-paced

1. Programming for ML

core
Production-Grade Python

Beyond notebooks. Typing, testing, packaging, async. ML engineers ship code that's read more than written.

SQL for Feature Engineering

Most training data starts as SQL against a warehouse.

2. ML Fundamentals

core
Classical ML

Linear/logistic regression, trees, random forests, gradient boosting. Solves 80% of production ML problems better than deep learning.

Model Evaluation

Metrics per problem type, cross-validation, calibration, cost-sensitive learning. Picking accuracy on an imbalanced dataset is the classic beginner error.

3. Deep Learning

core
PyTorch

The production default. Tensors, autograd, DataLoader, distributed training.

Transformers & Attention

The architecture behind LLMs, most SOTA vision models, and modern recommender systems.

4. LLMs & Generative AI

core
LLM Serving & Inference

vLLM, TensorRT-LLM, Ollama, llama.cpp. Quantization, batching, caching, throughput vs latency tradeoffs.

RAG & Vector DBs

Retrieval-augmented generation. Embeddings, vector search (pgvector, Weaviate, Pinecone, Qdrant), chunking strategies.

Fine-tuning

LoRA, QLoRA, full fine-tuning. When to fine-tune vs when to just prompt.

5. MLOps

core
Experiment Tracking

MLflow, Weights & Biases, Neptune. Reproducibility of runs is non-negotiable.

Model Registry & Serving

Register versioned models, promote through stages, serve behind a stable API.

Training Pipelines

Airflow, Kubeflow, Metaflow, Prefect. Orchestrate feature engineering, training, evaluation, deployment.

6. Feature Stores & Data

core
Feature Store Fundamentals

Feast, Tecton, Databricks Feature Store. Solve training/serving skew and feature reuse across models.

7. Model Serving

core
Serving Frameworks

TorchServe, Triton, KServe, Ray Serve, BentoML. Match to latency + throughput + hardware constraints.

8. Monitoring & Observability

core
Drift & Model Monitoring

Data drift, concept drift, performance decay. Evidently, WhyLabs, Fiddler, Arize. Models silently degrade — monitor them.

9. Distributed Training

recommended
Distributed PyTorch

DDP, FSDP, DeepSpeed. Scale beyond a single GPU/node.

10. Cloud ML Platforms

recommended
AWS SageMaker / GCP Vertex AI / Azure ML

Managed ML platforms — trade lock-in for less infra work.

11. Software Engineering Basics

core
Testing + CI/CD for ML

Unit test the code, integration test the pipeline, validate the model. GitHub Actions or GitLab CI for training + deploy.

12. Career & Reading

optional
Applied ML Blogs & Papers

Where ML engineers stay current.

Want a personalised version?

This roadmap is the same one our platform uses internally, but the logged-in version lets you tick off topics as you complete them, track a personalised First 90 Days plan, and see your AI-durability score against this role. All free.

Open the interactive roadmap →
Curated by WhatTNext Ai · methodology · all roadmaps · last updated 2026-07-29