AI Nexus (@ainexusgroup) profile picture

Where AI founders, researchers & builders meet. 1,500+ members · London · SF · Boston Meetups · Hackathons · Demo Nights https://teapawt.org/t.co/HE53nE1BE6

Joined February 2026
Photo by AI Nexus (@ainexusgroup)Photo by AI Nexus (@ainexusgroup)
My cold calling job is torture, so with zero prior experience, I spent my last few bucks making an AI ad for a YC startup, hoping it'd get me an interview... They left me on read. So I thought I’d post it here. (Name censored to avoid trouble)
🤖 Made with AI
2
2
5
2,478
AI Nexus retweeted
WE ARE THROWING THE BIGGEST API HACKATHON IN SF.
the time has come. api and cli products are the biggest things on tech twitter right now. so @nozomioai is spending $$$ to throw the biggest API hackathon in may. still accepting sponsors if you have cash or crazy prize ideas.
2
2
12
2,541
AI Nexus retweeted
the time has come. api and cli products are the biggest things on tech twitter right now. so @nozomioai is spending $$$ to throw the biggest API hackathon in may. still accepting sponsors if you have cash or crazy prize ideas.
19
6
4
156
16,459
Last Saturday. 6 cities. One mission: ship something that works. The Cactus x @GoogleDeepMind Global Hackathon brought together some of the sharpest builders we know.
1
1
43
From London to Singapore — hundreds of engineers racing the clock with FunctionGemma and Cactus Engine. Agentic AI. Local inference. Real constraints. Real demos. This is what AI Nexus is for 🖤 📍 London · San Francisco · Boston · LA · DC · Singapore
33
Cloud dev: "I'll just call the API" Local dev: already shipped Saturday. UCL East Campus. Cactus x Google DeepMind Hackathon. Build on Cactus Engine with FunctionGemma. No API keys. No rate limits. No excuses. 2 days. Register → link in bio
24
3 days until we build.
1
17
🛠️ Build agentic systems with FunctionGemma on Cactus Engine ⚡ Local-first AI infrastructure — edge + cloud, ship in one day Our last hackathon brought 700+ builders together. This one is bigger.
1
13
If you're an engineer, researcher, or founder who ships — join us Saturday. Register: ainexus.community
11
Alibaba just dropped Qwen3.5-397B. Open-weight, Apache 2.0.
1
12
397B total params but only 17B active at inference and sparse MoE + linear attention. 201 languages. 256K context. Native multimodality.
1
12
People are already running Q4 on Apple Silicon via MLX. You just need ~225GB of RAM. Ollama has it too. Probably the last big release before DeepSeek v4.
34