PLAN-101 Career OS
All Levels Estimated: 12 Weeks 12 Topics

AI Engineer Roadmap

Comprehensive 4-phase roadmap for AI Engineer covering 12 key topics with practical deliverables.

Original PLAN-101 Career Guide

The AI Engineer Roadmap provides an end-to-end curriculum for software engineers transitioning into applied artificial intelligence, generative AI, and large language model (LLM) engineering.

Unlike academic research paths centered purely on mathematical proofs, AI engineering focuses on integrating state-of-the-art foundation models into production software. You will master prompt orchestration, retrieval-augmented generation (RAG), vector embeddings, autonomous agent architectures, and model evaluation guardrails.

By completing this roadmap, you will bridge the gap between raw machine learning models and enterprise applications, learning how to build resilient, cost-effective, and latency-optimized AI systems.

Hands-on Projects and Acceptance Goals

  • Production RAG Engine with Hybrid Search: Document intelligence platform indexing PDF technical manuals into Qdrant/Pinecone with semantic vector search, BM25 keyword reranking, and citation generation. Tools: Python, LangChain/LlamaIndex, PostgreSQL, pgvector, FastAPI, React
  • Autonomous Multi-Tool Agent Gateway: Agentic assistant capable of dynamically selecting APIs (web search, calculator, SQL query engine), validating schemas, and retrying failed execution steps. Tools: Python, OpenAI/Gemini SDK, LiteLLM, Redis, LangSmith
  • Domain-Specific LLM Fine-Tuning & Evaluation Pipeline: Automated LoRA/QLoRA fine-tuning workflow with dataset curation, BLEU/ROUGE validation scoring, and LLM-as-a-judge regression test suites. Tools: PyTorch, Hugging Face Transformers, Unsloth, Weights & Biases

Frequently Asked Questions

Do I need a PhD or advanced math degree to become an AI Engineer? No. While deep theoretical ML research requires advanced mathematics, AI engineering primarily requires strong software engineering skills, API design, data processing, and an intuitive understanding of model behavior.

Should I learn LangChain, LlamaIndex, or raw SDKs first? We strongly recommend starting with raw SDKs (OpenAI, Gemini, Anthropic) to understand exact request-response formats and prompt structuring before adopting abstraction layers like LangChain.

What is the difference between an ML Engineer and an AI Engineer? ML Engineers traditionally train and deploy custom statistical and deep learning models from scratch. AI Engineers specialize in integrating, orchestrating, fine-tuning, and grounding modern foundation models into software products.

Curriculum Syllabus & Milestones

Phase 1: Foundations

Focus area covering 3 core topics in Foundations.

  1. LLM & AI Basics

    Learn what LLMs are, tokens, context windows and inference vs training.

    • Study primary concepts for: LLM & AI Basics
    • Actionable guide: Learn what LLMs are, tokens, context windows and inference vs training.
    • Build or test a practical mini-project applying LLM & AI Basics.
  2. Calling an LLM API

    Call OpenAI/Anthropic API directly and handle a streamed response.

    • Study primary concepts for: Calling an LLM API
    • Actionable guide: Call OpenAI/Anthropic API directly and handle a streamed response.
    • Build or test a practical mini-project applying Calling an LLM API.
  3. Prompt Engineering Basics

    Practice zero-shot, few-shot and structured-output prompting.

    • Study primary concepts for: Prompt Engineering Basics
    • Actionable guide: Practice zero-shot, few-shot and structured-output prompting.
    • Build or test a practical mini-project applying Prompt Engineering Basics.

Phase 2: Core Skills

Focus area covering 3 core topics in Core Skills.

  1. Embeddings & Vector Search

    Generate embeddings and do a similarity search over sample documents.

    • Study primary concepts for: Embeddings & Vector Search
    • Actionable guide: Generate embeddings and do a similarity search over sample documents.
    • Build or test a practical mini-project applying Embeddings & Vector Search.
  2. RAG Fundamentals

    Build a basic Retrieval-Augmented Generation pipeline over your own data.

    • Study primary concepts for: RAG Fundamentals
    • Actionable guide: Build a basic Retrieval-Augmented Generation pipeline over your own data.
    • Build or test a practical mini-project applying RAG Fundamentals.
  3. Function/Tool Calling

    Give a model a tool/function and handle its structured call correctly.

    • Study primary concepts for: Function/Tool Calling
    • Actionable guide: Give a model a tool/function and handle its structured call correctly.
    • Build or test a practical mini-project applying Function/Tool Calling.

Phase 3: Applied Practice

Focus area covering 3 core topics in Applied Practice.

  1. Building AI Agents

    Build a small agent that plans and calls 2-3 tools to complete a task.

    • Study primary concepts for: Building AI Agents
    • Actionable guide: Build a small agent that plans and calls 2-3 tools to complete a task.
    • Build or test a practical mini-project applying Building AI Agents.
  2. Open-source Models (Hugging Face)

    Run an open-source model locally or via Hugging Face Inference.

    • Study primary concepts for: Open-source Models (Hugging Face)
    • Actionable guide: Run an open-source model locally or via Hugging Face Inference.
    • Build or test a practical mini-project applying Open-source Models (Hugging Face).
  3. Multimodal Basics

    Use a vision or speech API to process an image or audio input.

    • Study primary concepts for: Multimodal Basics
    • Actionable guide: Use a vision or speech API to process an image or audio input.
    • Build or test a practical mini-project applying Multimodal Basics.

Phase 4: Career

Focus area covering 3 core topics in Career.

  1. Evaluation & Guardrails

    Add basic evaluation and safety checks to an AI feature.

    • Study primary concepts for: Evaluation & Guardrails
    • Actionable guide: Add basic evaluation and safety checks to an AI feature.
    • Build or test a practical mini-project applying Evaluation & Guardrails.
  2. Cost & Latency Optimization

    Measure and reduce token cost/latency for a real pipeline.

    • Study primary concepts for: Cost & Latency Optimization
    • Actionable guide: Measure and reduce token cost/latency for a real pipeline.
    • Build or test a practical mini-project applying Cost & Latency Optimization.
  3. Ship an AI Feature

    Build and deploy a complete AI-powered feature end to end.

    • Study primary concepts for: Ship an AI Feature
    • Actionable guide: Build and deploy a complete AI-powered feature end to end.
    • Build or test a practical mini-project applying Ship an AI Feature.

Related Career Paths & Cross-Skill Roadmaps

Expand your engineering versatility by exploring interconnected role roadmaps:

Data Science → Machine Learning → Backend Developer → 3-Year CSE Roadmap →

Track Your AI Engineer Roadmap Progress

Sign in with Google to check off completed topics, save code notes, and monitor your placement readiness score.

Start Tracking Roadmap