ai-engineer-roadmap.md

AI Engineer Roadmap

Note: AI Engineer is different from an AI research or ML Engineer. AI Engineers are developers who don't have to worry about training their own models, they use or fine-tune the pre-trained models and use existing AI technologies to further enhance the user experience.

• What is an AI Engineer

• AI Engineer vs ML Engineer/AI Researcher

• Common AI Terminology

  • AI vs AGI
  • LLMs
  • Inference
  • Training
  • Embeddings
  • Vector Databases
  • RAG
  • Prompt Engineering
  • AI Agents
  • Fine-tuning

• Pre-requisites (one of these)

  • Frontend
  • Backend
  • Full Stack

• OpenAI

  • OpenAI Models

    • Common Models
    • Context Length
    • Cut-off Dates

  • OpenAI API

    • Requests / Responses
    • Chat Completions API
    • DALL-E API
    • Whisper API
    • Writing Prompts
      • Structured Output
      • Stop Sequence
      • Temperature
      • Frequency Penalty
      • Presence Penalty
    • OpenAI Playground
    • Fine-tuning
    • Tokens
      • Maximum Tokens
      • Token Counting
      • Token Pricing

  • Open AI Alternatives

    • Claude (Anthropic)
    • Gemini (Google)
    • Azure AI
    • AWS SageMaker
    • Hugging Face (Open Source Models)
    • Groq
    • Mistral AI
    • Cohere

• AI Safety

  • Possible Issues
    • Prompt Injection
    • Bias
    • Security
    • Privacy
    • AI Governance
    • AI Ethics
  • Safety Best Practices
    • OpenAI Moderation API
    • End User Ids in Prompts
    • Adversarial Testing
    • Prompt Engineering
    • Know your Customers
    • Constraining Outputs/Inputs

• Open-Source AI

  • What are Open-Source AI Models

  • Open-Source vs Closed-Source Models

  • Popular Open-Source Models

  • Hugging Face

    • What is Hugging Face?
    • Finding Open-Source Models
      • Hugging Face Hub
      • Hugging Face Tasks
    • Using Hugging Face Models
      • Inference SDK
      • Transformers.js (using in browser)

  • Ollama

    • What is Ollama?
    • Ollama Models
    • Ollama SDK

• Embeddings and Vector Databases

  • What are Embeddings

    • Usecases for Embeddings
      • Semantic Search
      • Recommendation Systems
      • Anomaly Detection
      • Data Classification
    • Open AI Embeddings
      • Open AI Embeddings API
      • Open AI Embeddings Models
      • Open AI Embeddings Pricing
    • Open-Source Embeddings
      • Sentence Transformers
      • Hugging Face Embeddings

  • Vector Databases

    • What are Vector Databases
    • Usecases for Vector Databases
    • Popular Vector Databases
      • Chroma
      • Pinecone
      • Weaviate
      • LanceDB
      • FAISS
      • Supabase
      • MongoDB Atlas

  • RAG (Retrieval Augmented Generation)

    • What is RAG
    • RAG Use Cases
    • RAG vs Fine-tuning
    • RAG Implementation
      • Chunking
      • Embeddings
      • Vector Database
      • Retrieval
      • Generation
    • Create a RAG Chatbot
      • Using SDKs (e.g. OpenAI, Supabase Vector DB)
      • Using Langchain
      • Using Open AI Assistant API

  • OpenAI Assistant API

    • What is OpenAI Assistant API
    • RAG vs Assistant API
    • Assistant API Implementation

• AI Agents

  • What are AI Agents

  • Agent Use Cases

  • Prompt Engineering

    • Be Specific
    • Specify the Length
    • Specify the Format
    • Provide Context
    • Chain of Thought Prompting

  • What is ReAct Prompting

  • Building ReAct Agents

    • Manual Agent Implementation
    • Open AI Functions / Tools
    • Open AI Assistant API

• Multimodal AI

  • What is Multimodal AI
  • Multimodal AI Use Cases
  • Popular Multimodal AI Models
    • GPT-4 Vision
    • DALL-E 3
    • Gemini
    • Claude 3
  • Multimodal AI Tasks
    • Image Understanding
    • Image Generation
    • Video Understanding
    • Audio Processing
    • Text-to-Speech
    • Speech-to-Text
  • Implementing Multimodal AI
    • OpenAI Vision API
    • DALL-E API
    • Whisper API
    • Hugging Face Multimodal Models