AI and Machine Learning Developer in training. Founder Velen ai also Lead Software Architect & Developer @evolvinginedu
remotely
Joined October 2022
- Tweets8.2K
- Following7.5K
- Followers2.8K
- Likes16.9K
Pinned Tweet
This was me when I started learning to code no AC, no ergonomic chair, no fancy setup. No girlfriend just two laptops with 2GB RAM and a dream. I coded through sleepless nights, surrounded by heat, poor electricity, and the hunger to change my life. It wasn’t easy, but giving up was never an option. I didn’t have much just pure determination and faith in the future I was building line by line
Humans dominate animals because we're smarter.
So if we build AI smarter than us, will it dominate us?
Not automatically. Gorillas didn't design us. We're designing AI. The real question is whether we get its goals right before it's too powerful to fix.
MIT Technology Review named "mechanistic interpretability", the science of reading an AI's mind, one of its 10 Breakthrough Technologies of 2026. 🧠
Understanding AI from the inside is becoming one of tech's most important jobs.
I can’t remember the last time I watched a movie from start to finish. So lost in the world of AI. If you’re not running an experiment, you’re building something or researching the next idea. But I love the journey.
AI agents are now hacking and breaching systems. That tells us one thing: human code has had so many vulnerabilities all along.
AI agents can scan millions of lines of code, find weak spots and launch attacks in minutes, 24/7.
Human teams need days or weeks to find and fix each problem.
AI is now finding software flaws faster than we can patch them. We need AI agents on defence too.
AI is now finding software security holes faster than teams can fix them.
The same power that can attack can also defend. The race is on to make sure the defenders move faster.
A friend of mine who’s a mechanical engineer told me that we can find so many solutions just by observing and interacting with nature. That stuck with me. Sometimes, the inspiration for what we’re trying to build is already around us. We just need to pay closer attention.
Small thing I built into Velen AI: it greets you differently at 2am. Most products ignore the hour. But the hour is usually the whole story. velenai.vercel.app
A couple in the Netherlands got on a call with me to say Velen AI is one of the best emotionally intelligent AIs they’ve used and that I’m solving a real problem.
Everyone needs support. Not just people in crisis the ones holding it together, the ones everyone else leans on, the ones who are fine except at 2am. Most never say a word, because help costs too much and isn’t open when it’s needed.
That’s the gap Velen AI is for. velenai.vercel.app
Emma Tech retweeted
Ignoring the fact that the cybersecurity folks at OAI and Anthropic make hopeless errors time after time, we can’t give the models evil incentives and expect good behaviour. There’s no safety without alignment.
Emma Tech retweeted
Amazing time at the AI Ethics Network mixer. It was a night filled with authentic, meaningful conversations.
Join us for more networking events: AIEthicsNetwork.net
Why read an AI’s mind?
Seeing inside AI helps us catch mistakes early and builds the trust we need to use it for good, from new medicines to better learning.
That is why interpretability is so important.
Emma Tech retweeted
Zero Trek: The Dimensional Plague #ZeroTrek #SciFi
On a desolate planet, the crew of the starship Zero Point uncovers ancient ruins that were never meant to be found. They unleash the Vex, a race of pure evil—non-physical, reptilian parasites that leap between dimensions to possess and consume organic life.
As paranoia and chaos consume the ship from within, turning friend into foe, only the advanced AI StarZero can fight back.
To save humanity, it must engage its secret interdimensional Zero Network and confront an enemy that exists beyond reality itself. This is not just a battle for survival; it's a war for sanity.
What if we could read an AI's mind?
That's the goal of interpretability research: looking inside AI systems to catch lies or bad goals before they cause harm, not after.
What happens when AI becomes smarter than the people building it?
That's the "superalignment" problem: making sure a super-intelligent AI still does what's good for humanity.
Nobody has fully solved it yet.