AI Engineer Roadmap
You build, train, and deploy AI systems end-to-end
The full AI engineering stack - deep learning, LLMs, agents, and production deployment.
The route
- Week 1 - Python for ML: Your primary language - NumPy, pandas, scikit-learn are used daily
- Week 2 - AI Tools: Claude Code and OpenClaw multiply every week that follows - which is why they come early, not last
- Week 2 - Git & Version Control: Model versioning, experiment tracking, and ML-specific Git workflows
- Week 2 - Jupyter & Colab: Rapid prototyping and experiment iteration before production code
- Week 2 - SQL for Data: The metadata filter behind a retriever, the source of truth a RAG index was built from, and every text-to-SQL tool you will ship
- Week 3 - Foundations: Deep math fundamentals power every design decision you make as an AI engineer
- Week 3 - Statistics: Model evaluation, A/B testing, and statistical significance in experiments
- Week 4 - Probability Theory: Bayesian reasoning, distributions, and uncertainty quantification in models
- Week 4 - Linear Algebra: Embeddings, transformations, PCA - the math behind every neural network
- Week 5 - Calculus: Backpropagation, gradients, chain rule - how neural networks actually learn
- Week 5 - Optimization: Core skill - gradient descent, learning rates, loss surfaces, convergence
- Week 6 - Information Theory: Entropy, KL divergence, cross-entropy - needed before deep learning and LLMs
- Week 6 - ML Algorithms: Must-know for interviews - regression, trees, SVMs, ensembles, evaluation
- Week 7 - Deep Learning: Your bread and butter - CNNs, attention, transformers, training loops
- Week 8 - LLM: Build with LLMs - prompt engineering, RAG, fine-tuning, embedding pipelines
- Week 11 - Software Engineering for AI: Evaluation first, then production AI code - testing, architecture, system design, observability and cost
- Week 12 - Agentic AI & Multi-Agent Systems: Design agents with tool use, memory, planning and guardrails - and know when a deterministic workflow beats one
- Week 15 - MLOps & Deployment: Deploy and scale - Docker, monitoring, CI/CD, model registries
- Week 15 - ML Pipeline: Deployment-focused pipeline modules - model serving and monitoring
- Week 16 - AI on Cloud: Deploy your GenAI and agentic systems on AWS, GCP, or Azure - Bedrock, Vertex AI, AI Foundry, plus multi-cloud patterns
- Week 17 - Interview Q&A: Scenario-based interview prep - production problems that AI companies actually ask
- Week 17 - AI System Design: The AI system design round is its own interview, and the one that most often separates senior from mid. Sizing a retrieval index and defending a latency budget is not covered by the LLM or agents tracks