Today I want to officially introduce TECH TALK space by moseleydev A year ago, I started Tech Talk with 2–4 friends… just conversations, no audience, no hype. Just curiosity. Fast forward to today — we’ve hosted sessions with over 400 people in a single space. What started small is becoming something real. We run every 1st and 3rd Sunday — miss one, catch the next. Simple. But it’s no longer just a “space.” It’s where builders show up. Young techies ask real questions. Senior engineers share what actually matters in the industry . We’ve had people from Chevron, Ikeja Electric, edtech startups… and we’re just getting started. Next stop is KPMG, Moniepoint, Paystack, Google. Not because it sounds cool — but because we’re building a bridge between where you are and where you want to be. I’ve made real friends here. We share ideas, opportunities, direction. This isn’t a boring tech space at all. This is a growing ecosystem. We have active community on WhatsApp with my guy @r0ktech . If you want to speak, learn, or just listen in — you’re welcome, I’d love to connect. Let’s build one strong tech family here. #AI #buildingInPublic #Space #connect

What's your strongest skill? 👀 A. Building B. Marketing C. Sales D. Design E. Doom Scrolling 👀

I’m going to @VictorTrent31’s upcoming Space. Will you join too? x.com/i/spaces/1yxBeeeoNbLJN

Pls let’s share the space

How do you go about marketing a B2B product ? 🤔

9pm GMT*1 today x.com/i/spaces/1RJjppzZQddKw

can’t wait for the day I sit across from an interviewer and ask “so… why should I work for YOUR company?” that’s the goal 😹 #buildinpublic

Seems like we’ve hit the point where almost everyone is building the same thing. The next wave will look completely different. Hardware 👀 #buildinpublic

JOIN THIS SPACE NOW!! 😂😂 x.com/i/spaces/1dKrPEBjOROJX

Here are the repos we're talking about tonight 👇 🟢 Beginner • FreeCodeCamp — github.com/freeCodeCamp/free… • Docusaurus — github.com/facebook/docusaur… • OpenStreetMap — openstreetmap.org 🟡 Intermediate • Supabase — github.com/supabase/supabase • n8n — github.com/n8n-io/n8n • Langchain — github.com/langchain-ai/lang… • OpenTelemetry — github.com/open-telemetry/op… 🔴 Advanced • Kubernetes — github.com/kubernetes/kubern… • Node.js — github.com/nodejs/node • Apache Kafka — github.com/apache/kafka 🇳🇬 Africa • Flutterwave SDKs — github.com/Flutterwave • Paystack — github.com/PaystackHQ 🏛️ Linux Foundation • linuxfoundation.org/projects Pick one. Read the CONTRIBUTING.md. Find a good first issue. Ship your first PR. Your GitHub is your CV. Start filling it. #SundayTechTalks #OpenSource #DevCommunity

It’s weekend 👀 What are you building right now? Describe it in 5 words max link I’ll rate it out of 10 😀 #buildinpublic

It’s another weekend Are you building or resting ?

Honoured to be featured as a speaker at @Thrive_Links virtual webinar today! We’re diving into The Role of AI in Every Tech Niche — 7PM on Google Meet. See you there 🚀 Let’s talk AI🔥 Check comment for access link #buildinpublic

Set a reminder for my upcoming Space! x.com/i/spaces/1dKrPEBjOROJX

PHP is dead. C is dead. Java is dead. DSA is dead. Web development is dead. Software engineering is dead. You read posts like these every day, but the truth is that nothing is dead. You just need to get better at what you do. Keep learning, building, and growing.

DO YOU KNOW YOU CAN NOW RUN A 70B MODEL ON JUST 4GB OF VRAM?🤔🤯 No expensive hardware. No quantization. Just your regular machine. Meet AirLLM. Most people think running large LLMs requires a $10,000 GPU, 80GB of VRAM, and a cloud bill that hurts. AirLLM throws all of that out the window. Here’s the trick , instead of loading the entire model into memory at once, AirLLM loads ONE transformer layer at a time. Run it → unload it → load the next. Your disk does the heavy lifting. Your GPU just processes. The result? 🧵 70B parameter models on a consumer GPU 🧵 Full precision — no quality loss from quantization 🧵 Works with LLaMA, Mistral, and most HuggingFace models 🧵 Can even pull weights directly from S3 Getting started is dead simple: Just use pip install airllm The honest trade-off? It’s slower than in-memory inference because it reads layer by layer from disk. But if you’re experimenting, researching, or just don’t have the hardware ,speed vs access is a no-brainer. You take access every time. The barrier to running frontier-scale models just got a lot lower. No excuses. Fire it up → github.com/lyogavin/airllm RT if you know a dev who thinks they need better hardware 🔁