OpenTube

How I Build AI Projects From Scratch (Steal My Workflow)

Marina Wyss - AI & Machine Learning

27,300 views

Check out Supabase, I use them for all my projects! https://supabase.plug.dev/qmPbENX

Get my project scoping doc for free here: https://newsletter.marinawyss.com/ai-...

Timestamps ⏰
00:00 How I build AI projects from scratch
00:33 Why should you build AI portfolio projects?
01:46 What makes a good AI portfolio project?
03:34 Initial scoping
05:47 Working with AI coding assistants the RIGHT way
08:05 Setting up the project
08:47 Starting with an MVP
13:57 Further scoping
15:39 Working in phases
18:27 Evaluating AI projects
21:29 Monitoring AI projects
23:37 Deploying AI projects
24:41 Using the project to help get a job
---------------------------------------
🤖 Want to become an AI Engineer? Download my AI Engineering Skills Checklist here: https://newsletter.marinawyss.com/ai-...

💬 Interested in working with me 1:1? Learn more about my strategic advisory sessions: https://www.marinawyss.com/coaching?u...

🚀 Want to level up your AI/ML career? Join the AI/ML Career Launchpad community: https://launchpad.marinawyss.com/aiml...

🩷 Join the channel membership community for priority comment replies and early access to videos! @marinawyssai

📣 Invite me to speak: https://forms.gle/ZaJc83tiEQLAh9fC7

📖 Read the blog version here: https://www.marinawyss.com/blog/how-i...
---------------------------------------
🎥 Other videos you might like:

AI Engineering Roadmap: Projects that Get You Hired
AI Engineering Roadmap: Projects that Get ...

Your Projects Won't Get You Hired (Here's What Will)
Your Projects Won't Get You Hired (Here's ...
---------------------------------------
✉️ Contact
Business email: business@gratitudedriven.com
Instagram: Instagram: marina.wyss
LinkedIn: LinkedIn: marina-wyss
Twitter/X: https://x.com/iammarinawyss
TikTok: TikTok: marina_wyss
Leave me a comment here on YouTube!
---------------------------------------
❤️ My favorite things

Learn Python and AI engineering with JavaScript (perfect for software engineers): https://scrimba.com/?via=MarinaWyss

Learn AI system design: https://www.educative.io/courses/gene...

Learn ML and AI fundamentals hands-on: https://datacamp.pxf.io/6kDBPQ

My favorite technical books: https://www.amazon.com/shop/marinawys...
---------------------------------------
⚖️ Disclaimer
The views and opinions expressed in this video are my own and do not reflect the official policy or position of Twitch/Amazon or any other company I have worked for. All advice and insights shared here are based on my personal experiences and should be considered as such.

Thank you to Supabase for sponsoring this video!

This description may contain affiliate links. If you make a purchase I may make a small commission at no cost to you.

#aiengineering #aicoding #vibecoding

16 Comments

@feynman_QED 2 weeks ago

Fantastic, straight to the point and i love that template!
I don't see around people who teach so much in a short time, thank you Marina! 💪🏻

@NidaSea 2 weeks ago

Excited to give this a shot! Thank you, Marina!

@PuppetZombieMuppet 2 weeks ago

Solid framework. I think starting with a CLI MVP before any interface is underrated - it forces the core problem to work first.

@viveknidhi 11 days ago

Awesome project

@kanishcktewatia597 2 weeks ago

Solid advice. much appreciated. now i will never look at projects as "fun"

@jameskellam2980 2 weeks ago

There is a lot of great information here. I will have to keep coming back from time to time as I grow to learn more about the 1/2 or more of this video that was over my head
The take aways I got on the first pass that I can start implementing immediately: 1. Versioning the prompts. Excellent! I have downloaded various local AI models and used llama.cpp and ollama. Usually with the same prompt to eee how the flags change the outcomes and performance. Now I can start seeing how different prompts change the results too, and I will have a list of what works instead of a general idea. 2. The curveball about mvp. I tend to suffer from the "when all you have is a hammer, everything looks like a nail syndrome." So when I start playing deep enough to make an actual AI project I hope to remember to ask myself "do I really need AI for this, or would a program work better?" .

As always. Great video and I like the way you explain things.

@HilarioAntonioPerurena 9 days ago

Thanks a lot, it's very helpful. Just a feedback, when you try to download the scoping doc under safari, request the email to be downloaded, but the browser don't show the email box to be filled, under chrome is working good, just to mention it. Thanks again!

@johnhands6746 2 weeks ago

Hi Marina, awesome video! I really looked the output of the tests you did, comparing the models. How did you put this together? Thanks!

@TankaNafaka 12 days ago

Nice demo. Did you considered using conventional commits in git commits? Also what are your thoughts on githubs spec driven workflow in order to automate and validate users specs? 🤗

@hellomejellos 2 weeks ago

Appreciate the video Marina! Evaluations require data but I don't recall a video you've put together on how to gather that data. To what extent do you find silver set labeling with a more powerful model than you'd make accessible in your app to be okay, or would you just typically recommend projects that allow non-expert labeling (such as in your project example)?

@editor_atishay 2 weeks ago

Hey, it's Atishay. Just following up on the email I sent 20 days ago. I'm an editor and I genuinely believe I can improve your content. I ve also included a re-edit in that email to show you.