Machine Learning Engineer roadmap
A structured path to design, train, and deploy machine learning models and AI systems at scale, spanning mathematics, software engineering, core ML/DL, and MLOps.
- Topics
- 25
- Study time
- About 647 hours
- Domains
- 4
How the path is laid out
- Phase 1
Foundation
Math and programming fundamentals
- Phase 2
Intermediate
Core machine learning and deep learning
- Phase 3
Advanced
Modern architectures and production engineering
- Phase 4
Specialization
Scale, cloud, and system design
Syllabus: what you will learn
25 topics across 4 domains. Each has a lesson, an AI tutor and a quiz; a topic unlocks once the topics it builds on are mastered.
Mathematics3 topics · 100 hours
Linear Algebra
Vectors, matrices, eigendecomposition, SVD.
40 h
Probability & Statistics
Distributions, Bayes, estimation, hypothesis testing.
40 h
Optimization
Gradient descent, SGD, Adam, convexity, regularization.
Builds on: Multivariable Calculus, Linear Algebra
20 h
Software Engineering5 topics · 120 hours
NumPy & Pandas
Array computing and dataframe manipulation.
Builds on: Python Programming
25 h
SQL & Data Storage
Querying relational data, joins, window functions.
15 h
Git, Testing & Clean Code
Version control, pytest, code review practices.
Builds on: Python Programming
15 h
Data Structures & Algorithms
Complexity, core structures, interview-level algorithms.
Builds on: Python Programming
50 h
Building APIs
REST APIs with FastAPI, async, validation.
Builds on: Python Programming
15 h
Machine Learning9 topics · 230 hours
Supervised Learning
Linear/logistic regression, trees, gradient boosting, SVMs.
Builds on: Classical Machine Learning, Linear Algebra
35 h
Unsupervised Learning
Clustering, PCA, dimensionality reduction, anomaly detection.
Builds on: Classical Machine Learning, Linear Algebra
20 h
Model Evaluation & Validation
Cross-validation, metrics, bias-variance, leakage.
Builds on: Supervised Learning
15 h
Feature Engineering
Encoding, scaling, feature selection, handling missing data.
Builds on: NumPy & Pandas, Supervised Learning
15 h
Neural Network Fundamentals
Perceptrons, backprop, activations, initialization, regularization.
Builds on: Deep Learning, Optimization
25 h
PyTorch
Tensors, autograd, modules, data loaders, training loops.
Builds on: Neural Network Fundamentals, Python Programming
30 h
Convolutional Networks & Vision
Convolutions, ResNets, transfer learning for vision.
Builds on: PyTorch
25 h
Transformers & Attention
Self-attention, encoder/decoder, tokenization, fine-tuning.
Builds on: PyTorch
30 h
LLMs & Generative AI Systems
Fine-tuning (LoRA), RAG, embeddings, evaluation of LLM apps.
Builds on: Transformers & Attention
35 h
MLOps & Infrastructure8 topics · 197 hours
Docker & Kubernetes
Containerization and orchestration of ML workloads.
Builds on: Git, Testing & Clean Code
30 h
Experiment Tracking & Model Registry
Versioning data, models, and experiments.
Builds on: Model Evaluation & Validation, Git, Testing & Clean Code
12 h
Data & Training Pipelines
Orchestrating ETL and training with Airflow, CI/CD for ML.
Builds on: SQL & Data Storage, Docker & Kubernetes, Feature Engineering
25 h
Model Serving & Inference Optimization
Batch vs online inference, ONNX, quantization, Triton, vLLM.
Builds on: Building APIs, Docker & Kubernetes, PyTorch
30 h
Monitoring & Drift Detection
Data/concept drift, performance monitoring, alerting.
Builds on: Model Serving & Inference Optimization, Probability & Statistics
15 h
Distributed Training & GPUs
Data/model parallelism, DDP, FSDP, mixed precision.
Builds on: PyTorch, Docker & Kubernetes
30 h
Cloud ML Platforms
AWS SageMaker, GCP Vertex AI, managed compute and storage.
Builds on: Docker & Kubernetes
25 h
ML System Design
End-to-end design of recommenders, search, ranking at scale.
Builds on: Model Serving & Inference Optimization, Data & Training Pipelines, Monitoring & Drift Detection, Data Structures & Algorithms
30 h
Find out how much of Machine Learning Engineer you already know
The placement test takes about five minutes and marks the topics you can skip.
Start this roadmap