The AI Engineer Path – Scrimba #

https://www.coursera.org/specializations/ai-engineering#courses

Intro to AI Engineering (104 min) #

• Welcome to The AI Engineer Path!
• AI Engineering basics
• The code so far
• Polygon API sign-up & key
• Get an OpenAI API Key
• Overview of how the API works
• An API call: OpenAI dependency
• An API call: Instance and model
• An API call: The messages array
• A quick word about models
• Prompt Engineering and a challenge
• Adding AI to the App
• Tokens
• The OpenAI Playground
• Temperature
• The “Few Shot” Approach
• Adding Examples
• Stop Sequence
• Frequency and Presence Penalties
• Fine-tuning
• Creating Images with the DALL·E 3 API
• Intro to AI Safety
• Safety Best Practices
• Solo Project - PollyGlot
• You made it!

Deployment (50 min) #

• Learn secure & robust deployment strategies
• Create a Cloudflare worker
• Connect your worker to OpenAI
• Update client side data fetching
• Handle CORS and preflight requests
• OpenAI API requests & responses
• Create an AI Gateway
• Error handling
• Create & deploy the Polygon API worker
• Fetch the stock data
• Download files and push to GitHub
• Deploy your site with Cloudflare Pages
• Custom domains with Cloudflare
• Recap & next steps

Open-source Models (33 min) #

• Open source vs closed source
• Intro To HuggingFace.js Inference
• Text To Speech With HuggingFace.js Inference
• Transforming Images with HuggingFace.js Inference
• AI Models In The Browser With Transformers.js
• Download and Run AI Models on Your Computer with Ollama
• Section Recap

Embeddings and Vector Databases (94 min) #

• Your next big step in AI engineering
• What are embeddings?
• Set up environment variables
• Create an embedding
• Challenge: Pair text with embedding
• Vector databases
• Set up your vector database
• Store vector embeddings
• Semantic search
• Query embeddings using similarity search
• Create a conversational response using OpenAI
• Chunking text from documents
• Challenge: Split text, get vectors, insert into Supabase
• Error handling
• Query database and manage multiple matches
• AI chatbot proof of concept
• Retrieval-augmented generation (RAG)
• Solo Project: PopChoice

Agents (117 min) #

• AI Agent Intro
• Prompt Engineering 101
• Control Response Formats
• Zooming Out
• Agent Setup
• Introduction to ReAct prompting
• Build action functions
• Write ReAct prompt - part 1 - planning
• ReAct Agent - part 2 - ReAct prompt
• ReAct Agent - part 3 - how does the “loop” work?
• ReAct Agent - part 4 - code setup
• ReAct Agent - part 5 - Plan for parsing the response
• ReAct Agent - part 6 - Parsing the Action
• ReAct Agent - part 7 - Calling the function
• ReAct Agent - part 8 - Housekeeping
• ReAct Agent - part 9 - Finally! The loop!
• OpenAI Functions Agent - part 1 - Intro
• OpenAI Functions Agent - part 2 - Demo day
• OpenAI Functions Agent - part 3 - Tools
• OpenAI Functions Agent - Part 4 - Loop Logic
• OpenAI Functions Agent - Part 5 - Setup Challenge
• OpenAI Functions Agent - Part 6 - Tool Calls
• OpenAI Functions Agent - Part 7 - Pushing to messages
• OpenAI Functions Agent - Part 8 - Adding arguments
• OpenAI Functions Agent - Part 9 - Automatic function calls
• Adding UI to agent - proof of concept
• Solo Project - AI Travel Agent
• Nice work!

Multimodality (62 min) #

• Introduction
• Generate original images from a text prompt
• Response formats
• Prompting for image generation
• Size, quality and style
• Editing images
• Image generation challenge
• Image generation challenge solution
• GPT-4 with Vision - Part 1
• GPT-4 with Vision - Part 2
• Image generation & Vision recap

OpenAI’s Assistants API (30 min) #

• Introducing the Assistants API
• How OpenAI Assistants work
• Create an Assistant
• Create a thread and messages
• Running an Assistant
• Bring it all together
• More to explore