I’ve worked on the product side of building genAI features since last September, which naturally came with a lot of “drinking from the fire hose” w.r.t. keeping up with AI news. It was this May when I decided to take the next step and dive into learning the technical side of genAI — after all, my CS degree AI/ML specialization does count for something! After a whirlwind of exploration, I’ve compiled and would like to share a resource masterlist of the resources I’ve found handy in my learning journey.
As a preview to what will be covered below, here’s what I’ve done in the last few months:
- kept myself updated with the latest AI news via podcasts and newsletters
- attended events up and down SF and South Bay to really get tapped into what’s happening (and met several power players in the space, including LangChain founder Harrison Chase!)
- completed a Udacity nanodegree in GenAI, including getting hands on with finetuning GPT-2 and building an agentic system with RAG
- completed all the GenAI paths in Google Cloud Skills Boost, familiarizing myself with Gemini and Vertex AI
- taken various short courses on DeepLearning.ai to upskill on LangChain and other frameworks
- built touchgrass, an AI local guide to recommend and answer questions about places on Google Maps (here’s the Github repo and demo video below) to get hands-on with the knowledge I’ve gained
My hope is that my experience with these courses, newsletters, and event hubs will be helpful to anyone looking to learn more about GenAI, whether that means dipping their toes in or jumping into the deep end. Without further ado, these are the resources I’ve gathered!
Table of Contents
Keeping Up with News
∘ Email Newsletters
∘ Blogs
∘ Podcasts
∘ Events
Upskill or Build a Foundation with Courses
∘ DeepLearning.ai
∘ Google Cloud Skills Boost
∘ Udacity GenAI Nanodegree
Keeping Up with News
Email Newsletters
TLDR AI: “Get smarter about AI in 5 minutes:”
- Length: Medium
- Audience: Technical and Business (news, papers, and repos)
- Newsletter Structure:
— Headlines & Launches
— Research & Innovation
— Engineering & Resources
— Miscellaneous
— Quick Links
The Batch @ DeepLearning.AI: “What Matters in AI Right Now”
- Length: long
- Audience: Technical and Business
- Newsletter Structure:
— Letter from Andrew Ng
— Featured DeepLearning.AI course
— News (a dedicated section per headline)
AlphaSignal: “A newsletter for developers by developers”
- Length: Medium
- Audience: Technical (news, papers, models, and repos)
- Newsletter Structure:
— One news highlight and a list of trending topics
— Deeper dive into the top news highlight
— A deep linked repeat of the trending topics
Ben’s Bites: “Simple, bite-sized education designed to boost your AI knowledge”
- Length: Short
- Audience: Business (with a focus on promoting trending or up and coming AI tools)
- Newsletter Structure:
— 3–5 top headlines
— AI tool shoutouts
— Occasional deeper dive into a headline
Blogs
AI Snake Oil: What Artificial Intelligence Can Do, What It Can’t, and How to Tell the Difference
- Review Pending
- Enjoyed the GPT-4 and professional benchmarks: the wrong answer to the wrong question article
Podcasts
AI Daily Brief: Short, daily show on the most important news in AI
- Length: ~15 minutes per episode
- Audience: Business
- Structure:
— Headlines Edition (5 minutes of all the headlines)
— Main AI Breakdown (10 minutes deep diving into one headline)
Events
- Local events in your city w/a focus on tech and business
- Sub to specific calendars (e.g. Generative AI San Francisco and Bay Area · Events Calendar)
- Similar to Lu.ma (lots of events cross-listed between the two) but specifically for AI events
- Limited to:
— SF & Bay Area
— NYC
— London
— Seattle
— Remote
Upskill or Build a Foundation with Courses
The courses below are listed in ascending (least to most) order of time commitment/involvement.
DeepLearning.ai
Overview: Short lessons with interactive Jupyter Lab notebooks alongside video tutorials
Pros: Helpful to get a quick overview of developer frameworks and tools (e.g. LangChain, CrewAI)
Cons: Not the most conducive to hands-on learning if you’re just starting out since the Jupyter notebook is already filled out for you and you just need to run through the cells
Google Cloud Skills Boost
Overview:
- Conceptual knowledge videos
- “Regular” labs with Jupyter notebooks already filled out for you
- “Challenge labs” with Jupyter notebooks that have blank TODO sections requiring some hands-on work
Pros: Good coverage of both concepts and code
Cons: most of the time you just needed to remember where to copy some code from the regular labs and paste into the TODO sections in the challenge lab. Some brainwork needed, but not as involved as it could be
Learning paths:
For developers with some knowledge of AI:
- Advanced: Generative AI for Developers Learning Path
- The Develop GenAI Apps with Gemini and Streamlit lesson within the Intermediate: Gemini for Google Cloud Learning Path is also helpful if you want to build a prototype with Gemini + Streamlit (though in my opinion, not as helpful as reading the documentation directly)
For Introductory learners/non-technical people:
- Start off with Beginner: Introduction to Generative AI Learning Path
- Then, watch the videos in modules 1–4, 7 (and 5 if you’re working with images) in the Advanced: Generative AI for Developers Learning Path. These short videos will cover:
— fundamentals in technical advancements up to the present with genAI
— Current techniques (vector embeddings and vector search) that are key to many RAG systems and also being used in regular search
For agent-specific topics:
- Google Startup School: GenAI Building AI agents: The curriculum here mirrors a lot of what you see in the Advanced GenAI Learning Path, and may be a good way to follow along with an actual “class” while you complete the labs.
Tip: the Cloud Skills Boost learning leaderboard was a fun way to motivate myself to learn each week! I made it to the highest rank (Diamond) just in time to finish all the GenAI related lessons.
Udacity GenAI Nanodegree
Overview: A 4 month program with lectures, labs with solution walkthroughs, and 4 hands-on projects that you submit to graders (no solution key).
Pros:
- The most in-depth conceptual learning and hands-on coding I’ve had in a genAI curriculum
- extensive lectures on the mathematical and foundational concepts
- freeform projects that allow you to apply what you learn without an “answer key” fallback
- Udacity AI is a powerful and convenient AI tutor
— a chatbot powered by OpenAI
— context-aware of the material you’re learning in your course
— conveniently located as a slide-out panel directly on top of my Udacity Classroom interface (no need to switch tabs to ChatGPT/Gemini/Claude)
- Active student forum with human instructor support
— I did not regularly use this feature, but it came in handy when I was looking for clarification on my first project’s instructions
Cons:
- some of the projects used ‘outdated’ code (e.g. using openai.Completion.create instead of client = OpenAI() and client.chat.completions.create)
— However, this is almost inevitable for a larger, more involved program that can’t be as easily updated as a more modular course (e.g. with DeepLearning.ai or Cloud Skills Boost) and isn’t too big of a problem if you’re using Udacity’s own workspaces (Jupyter notebooks hosted in their own environment)
Syllabus and Course Content:
Generative AI Fundamentals
- Project 1: Lightweight Fine-Tuning a Foundation Model
— Finetuned GPT-2 with Hugging Face and Pytorch tools
Large Language Models (LLMs) and Text Generation
- Project 2: Custom Chatbot
— Built a chatbot with a RAG system using OpenAI and semantic search
Computer Vision and Generative AI
- Project 3: AI Photo Editing with Inpainting
— Used the Segment Anything Model (SAM) and Stable Diffusion to replace parts of images with AI-generated content
Building Generative AI Solutions
- Project 4: Personalized Real Estate Agent
— Built a real estate agent with a RAG system using the OpenAI API, vector databases and semantic search, and the LangChain framework.
Tip: If you decide to try and condense the 4 month timeline while also balancing a full time job, you can finish within a month or two
And that’s all for my learning journey as of the end of August! I’ll be keeping an up-to-date (and if I were to be honest, a more easily formatted) version of this masterlist on this Notion page. I’m always on the lookout for more ways to keep up to date or learn, so feel free to comment any resources you’ve found useful and I might add them to the list!
