Heyy, Im Kedasha! And I'm a Senior Developer Educator (officially promoted!)! 💕💃🏽
--
7 years ago I set out on a seemingly impossible mission - to dramatically increase my income. At the time, I was earning $28k/year and struggling.
I learned about coding and software development from Instagram and fully committed to the dream - I saw a way to make that dramatic change and within 10 months, I went from $28K to over $90K, hired as an Associate Software Engineer at a fortune 100 company.
Life changing.
Within 2 years, I started sharing content online, writing blog posts, speaking at conferences, doing podcast appearances, got promoted to Software Engineer and landed a dream role at GitHub as a Developer Advocate.
Since then, I've created countless amounts of technical content, spoken on the largest developer stages in the world with Satya Nadella (Msft CEO) and Thomas Dhomke (former GitHub CEO), delivered so many workshops and talks and most of all (my pride and joy) created targeted content for Developers new to GitHub - GitHub for Beginners - reaching over 10 million developers worldwide.
My personal channels have continued to grow and make an impact on the community - reaching over 800K people worldwide.
It has been a journey. A fun-filled, high energy, work intensive journey and I wouldn't change a thing (except some overworking, haha).
I am incredibly proud of myself.
Thank you for being on this journey with me and cheers to the next 5 years of dramatic change.
#ai #developer #blackintech
There's a Wall of Shame for vibe-coded apps and it just went viral on Hacker News.
Exposed user data. Passwords stored in plain text. URL bar account takeovers.
The scary part? Every single one of those devs thought their app was secure. The Coding Agent said it was fine. So they shipped.
Here's how you make sure that's never you:
🔍 Search your entire codebase for any API key or secret that isn't in an environment variable.
🔓 Try to access every page and endpoint without being logged in.
⚡ Drop a single quote into every input field and watch what happens.
That's it. That catches 90% of what's on that list.
Don't be on it.
#vibecoding #ai #saas
Can we talk about something the AI agent hype cycle keeps skipping? 💭
Everyone's building agents to do more. More tools, more access, more autonomy.
But almost nobody's stopping to ask: what happens when your agent does something you didn't actually tell it to do?
Because that's not a hypothetical. It's already happening. Agents are writing code, hitting APIs, and touching databases and most of them are running with zero rules about what's off-limits.
If a new hire showed up and you handed them every password you own, your team (and you) would lose their minds!
When we do it with AI, somehow we call it progress. 🫠
The next real skill in this space isn't building flashier agents, it's knowing how to put guardrails around them.
Stick around, that's what I'm getting into next.
I'd love to know your thoughts - have you been adding guardrails to your agents?
#aiagents #aisecurity #vibecoding
You don't have to level up to contribute to open source. You level up by contributing to open source.
Not sure how to get started? Check out our latest GitHub for Beginners episode.
If you've ever pasted a work doc into ChatGPT to "just clean it up real quick" - congrats, you've participated in Shadow AI.
I learned this term at Red Hat Summit and now I can't unsee it.
Shadow AI is the AI version of shadow IT where employees use unapproved AI tools to do their job. It feels productive and it often is. But the risks are real:
→ Customer data exposed in tools that learn from your inputs
→ PII compliance violations
→ HIPAA risk for anyone in healthcare
→ Sensitive info sitting in someone else's cloud, indefinitely
What companies should be looking at: confidential computing, the model sees what it needs, but the data stays encrypted end-to-end. Even if someone accesses it, it's useless.
And my favorite tip from the conversation: before using any AI tool at work, ask the tool itself what data it sends back to its provider. It'll tell you. Use that to make an informed call.
Shadow AI isn't going away. The smart move is to name it, understand it, and put guardrails around it instead of pretending like it isn't happening.
Had you heard this term before? Let's chat in the comments👇🏽
This video is sponsored by @red_hat
#aiagents #shadowai #redhatsummit
Nobody's talking about this yet, but agent memory is about to be the next big AI discipline. 🧠
I just spent the week with engineers and execs at Oracle Developers in Palo Alto and learned more about agent memory than I knew existed.
Here's an introduction:
Agent memory = stored info an AI agent can retrieve across sessions. Unlike a context window, it's persistent and scoped to the user.
There are 4 types every builder should know:
🟢 Working memory - what the agent has access to right now (your current chat or coding session)
🩷 Procedural memory - how the agent does things (skill.md, instructions, rules files)
🩵 Semantic memory - facts the agent learns (your fave smoothie, your brand guide)
🟡 Episodic memory - specific events the agent can recall (past sessions, failed attempts)
The most interesting part to me is that not every agent needs every memory type. That's where Agent Memory Engineers will come in.
I have a full bloggy coming soon. In between time comment MEMORY and I'll send you a linky to read more about agent memory!
This video is sponsored by @oracle
#aiagents #agentmemory #aiengineering
If you've ever shipped something built with AI and gotten that little voice that says "…is this actually safe?" This is for you.
Here's the truth: AI tools are optimized to make things work. Not to make them safe.
So when you vibe code an app, the default output usually has:
→ API keys exposed on the client
→ Endpoints with zero authentication
→ Database queries vulnerable to injection from the browser
You're not doing anything wrong. This is just what AI does automatically.
The good news? The same 10 holes show up in nearly every vibe-coded app I've audited and most of them take 2 minutes or less to fix.
I made you a checklist with all 10 + the fix for each one.
Comment "checklist" below and I'll DM you the link 💌
#ai #vibecoding #appsecurity
I have a tendency to work incredibly too much. But, Im not just working for myself.
I work hard for these two - my hearts walking around earth - mom & grabdma.
I've had the best time enjoying them in Jamaica. I'm not an emotional girlie, so I don't always say what Im feeling, but these two, are the reason I do everything. Love them so much.
Happy mother's day to all who celebrate 💕
#mothersday #immigrant
If you're vibe coding you need to understand these 3 things before your project breaks:
→ Vector embeddings: how AI understands your data
→ Vector search: why AI pulls the wrong answer even when the right one exists
→ Vector databases: the foundation that makes or breaks everything
While building an AI Receptionist for my brother, I leaned heavily on the resources in MongoDB’s AI Learning Hub.
Comment “embeddings” and I’ll send you the courses I've been taking.
Now, go learn this stuff before your project teaches you the hard way.
This video is sponsored by @mongodb
#mongodb #ai #vibecoding #vectorsearch #techeducation
If you're feeling overwhelmed by AI right now, you're not alone! It's literally impossible to keep up. There’s a new tool every week that’s a "game changer” and it's exhausting.
The only thing that helps me cut through the noise is to choose one problem and go deep on it at any given time.
At this moment that’s agent memory - and this came directly from building my brother’s AI Receptionist.
Since LLMs are stateless by default, a key question is where do you store the agent’s memory? A database/. Vector store? In a file system?
I came across this super interesting notebook from the dev team at Oracle that talks about AI agent memory - comment MEMORY and I'll send you the linky.
In between time, let me know your thoughts - what one problem will you choose and go deep on?
This video is sponsored by @oracle
#aiagents #aiengineering
