David (@defemisback) header
David (@defemisback) profile picture

Performance & Software Engineering | C++ Go PostgreSQL Kotlin | prev. @Gamestop @Amazon @Gate

Building Web3
Joined March 2021
I am unable to claim any of the trading fees from $ApeBrain Took some time to understand how the whole pump.fun thing works and it appears the dev abused a vaulnorability to keep the trading fees to himself.
5
3
683
unpopular opinion: gemini 3.1 pro is a pretty great model most current baseline models are actually perfectly fine for 99% of everyday work. not everyone needs a state-of-the-art coding model to get things done.
2
1
16
1,197
My GPT-6 Astra 3D builds just passed 15M views on X. 🚀 The question I get most: “How are you getting this level of quality?” Here’s exactly how. Full guide 👇
8
1,261
Open ecosystems are the foundation of a healthy AI industry. 🐟 Proud to see @SakanaAILabs standing alongside global tech leaders in support of the open-weights letter. Collective intelligence is the future. 🚀
12
963
Get your own ai[dot]studio domain for free with any site or app you build. Free hosting. Free domains. Free building. The next wave of launches is giving you everything you need to turn your ideas into reality.
5
1,300
Introducing 3 new Gemini models. ⚡ • 3.6 Flash, faster, cheaper, sharper at coding • 3.5 Flash-Lite, fastest in the series at 350 tokens/s • 3.5 Flash Cyber, autonomously finds and fixes security vulnerabilities Available today in AIS and the API. 🚀
9
1,133
I used Fable 5 to bring Command & Conquer: Red Alert 2 + Yuri’s Revenge to iPhone and iPad. EA never released the source code, so there was nothing to port. Gemini 3.6 Flash play-tested it for hours, while GPT-5.6 Sol rebuilt the skirmish AI. Open-sourcing it all below. 👇
5
225
I just built the first fully operational recurrent brain of an ape you can run locally, open up, experiment on and watch process information in real time. Inside the the ape's brain: → 144 interconnected recurrent units → Persistent internal state & memory → Attention and fixation mechanisms → Valuation & decision-making → Threat / caution responses → Action selection → Live neural activation visualizer → Memory inspection → Deterministic brain-state replay → Pause, step through & inspect its thoughts → Remove memory, recurrence or threat response and observe how behavior changes → Feed it sequential environments and watch the same brain adapt its behavior over time → Fully inspectable source - no hidden model or API Markets are simply the first environment we've plugged into it. ApeBrain observes a stream of information, forms internal state, remembers previous signals, focuses its attention and eventually acts. The interesting part isn't what it trades. It's that you can watch the entire decision process happen from the inside. Fork the brain. Break it. Change its environment. Give it new senses. See what happens. Everything is open source.
13
3
14
6,486
This took just 3 prompts with SeedDance 2.5 in @runwayml. The leap AI video has made in such a short time is insane.
2
701
I used GPT-6 Astra to get AAA PC games running on my iPad mini! Skyrim, Batman: Arkham City, Hades, and Age of Empires II. The actual Windows versions, running locally with touch controls. No streaming. Demo below and I'll share a repo and setup guides soon!
3
727
FLUX.2 turns live Google Earth and Pinterest screens into ink drawings, clay models, felt maps, and miniature worlds. 🤯 The demo runs at original speed, no optical flow or artificial speed-up. Inspired by @Ryan__Stephen.
3
324
With Astra and Fable 5.1, we officially have drop-in remote workers. 🤖 It still takes some setup, but once dialed in, it’s almost seamless. I haven’t touched my computer in days, yet I’m more productive than ever.p
3
196
I wireheaded a fly and forced it to doomscroll FlyTok. Its dopamine neurons are measured and artificially enhanced for maximum enjoyment. The goal is simple: Create a fly happier than every other fly combined. 🪰
5
304
Gave my mom a very simple agent setup. She named him Morgan. He just closed a $600 ad deal for her site. People are massively overthinking agents. The Bitter Lesson applies here too: just ask the model, let it spawn what it needs, and get out of the way.
3
153
If you’re a bootstrapped or lightly funded AI founder with a product getting 500+ new signups per day, DM me. I have something for you. 👀 Please don’t DM if you’re below the 500 signups/day threshold.
4
437
Silicon Valley once dismissed Japan’s System Integration (SI) culture as an unscalable consultant trap. AI changes the equation. Writing the system is no longer the scarce work. Integrating it into real workflows, businesses, and organizations is.
3
146
🚀 AI Studio mobile app update: After 800K+ pre-orders across 168+ countries, Google says it’s taking a different approach, instead of launching another standalone app, it’s partnering with the Gemini app team to let apps emerge naturally through everyday conversations.
2
142
Is anyone experimenting with Jev for scalable oversight? It seems like it could be useful for evaluating whether a model is actually behaving in an aligned way. 👀
1
147