Must Read
Book
Designing Machine Learning Systems
Build intuition for production ML trade-offs across data, deployment, monitoring, and system design.
Intermediate4–6 hours
Curated by an ML Engineering Leader
Learn the engineering behind production AI.
A practical learning hub for engineers building ML, LLM, agentic, and data-intensive systems. Follow a roadmap, explore a topic, build a project, or use a growth path for mentoring.
Choose how you want to learn
Different people enter from different starting points. The same content can support self-learning, mentoring, and interview preparation.
01 · Learn
Follow a Roadmap
Move from fundamentals to production ML systems and architecture.
02 · ExploreExplore by Topic
Jump directly into LLMs, evaluation, MLOps, data systems, agents, and more.
03 · BuildPractice by Building
Use projects and system-design prompts to convert reading into engineering skill.
04 · GrowUse for Mentoring
Map role expectations, identify skill gaps, and build focused development plans.
ML Engineering Roadmap
A role-oriented progression that emphasizes engineering judgment, not just model knowledge.
Stage 01
Foundations
Python, SQL, probability, statistics, linear algebra, software engineering basics.
Stage 02
Applied Machine Learning
Supervised learning, feature engineering, experimentation, metrics, error analysis.
Stage 03
ML Systems
Training pipelines, model serving, data validation, testing, batch and online inference.
Stage 04
Production ML
MLOps, observability, drift, reliability, scalability, cost, incident response.
Stage 05
LLM & GenAI Engineering
Transformers, RAG, structured generation, fine-tuning, evaluation, inference systems.
Stage 06
Agentic Systems
Tool use, orchestration, memory, planning, hybrid workflows, reliability, trajectory evaluation.
Stage 07
Architecture & Leadership
Platform design, trade-offs, technical strategy, architecture reviews, mentoring, team execution.
Curated resources
Each recommendation includes the reason to read it, expected level, and how it connects to production work.
Deep Dive
Guide
Production LLM Evaluation
Learn how to combine functional metrics, golden datasets, error taxonomies, and model-based judging.
Advanced60 min
Recommended
Learning Path
Reliable Agentic Systems
Focus on tool-use, deterministic guardrails, trajectory correctness, observability, and recovery.
Advanced3 hours
Must Read
Topic
RAG: From Demo to Production
Understand retrieval quality, chunking, reranking, grounding, failure analysis, and evaluation.
Intermediate90 min
Mentor Pick
Playbook
Senior → Staff ML Engineer
A capability framework for moving from local execution to architecture, leverage, and technical leadership.
Senior45 min
Hands-on
Project
Build a Model Serving Benchmark
Compare latency, throughput, batching, autoscaling, and cost under realistic inference workloads.
IntermediateWeekend
No resources match your current search and filter.
Role-based learning paths
Opinionated sequences that answer a more practical question: “What should I learn next for the role I want?”
Production LLM Engineer
For engineers moving from LLM prototypes into reliable production systems.
TransformersRAGEvaluationServingObservability
Agentic AI Engineer
For engineers building tool-using agents and hybrid deterministic + agent workflows.
Tool CallingPlanningMemoryEvaluationReliability
Senior ML Engineer
For experienced engineers growing into broader system ownership and architecture.
Distributed MLPlatformsReliabilityCostArchitecture
Mentoring toolkit
Use the same learning hub as a lightweight framework for career conversations and growth planning.
ML Engineer → Senior ML Engineer
Assess the engineer across technical depth, production ownership, system design, debugging, and influence.
Can independently own an ML system from data to production.
Understands reliability, observability, cost, and operational trade-offs.
Can lead technical design for a multi-component ML feature.
Improves team standards through reviews, tooling, or mentorship.
Turn gaps into a focused plan
1. Assess: choose the target role and identify 2–3 capability gaps.
2. Learn: assign a small number of must-read resources from the hub.
3. Build: pair the reading with a project or design exercise.
4. Discuss: use architecture and failure-analysis questions in 1:1s.
5. Review: revisit the capability matrix after real production experience.
Build depth, not just a bookmark collection.
Every topic should connect concepts → engineering judgment → resources → hands-on practice.