Rami's Readings #142 - 🌟 Qwen3.5-Medium
Sharing Level 1 Innovation Governance Program certification and the latest on AI, LLMs, Qwen3.5, Olmo, Bot Traffic, Japanese Web Design, and more.
Welcome to Rami’s Readings #142 - a weekly digest of interesting articles, papers, videos, and X threads from my various sources across the Internet. Expect a list of reads covering AI, technology, business, culture, fashion, travel, and more. Learn about what I do at ramisayar.com/about.
👋🏼 Welcome New Subscribers
Hello! A hearty thank you for subscribing to Rami's Readings! There are quite a few new subscribers this week, thanks to recommendations from The AI Ethics Brief, Product Byte, and The VC Corner. I am thrilled to have you on board! In this newsletter, I curate the best papers, tweets, and articles I have read during the week focusing on LLMs, AI, economics, business, and technology news. You can learn more about me on my website.
🎙️ Podcast with Framework Venture Partners
I recently sat down with Mikhail Karmali from Framework Venture Partners for a wide-ranging conversation recorded at my kitchen table. 🙂 We discussed how the LLM era is compressing traditional SaaS models faster than many appreciate. The cost and complexity of building software has fallen dramatically, yet the enterprise buy-vs-build bar has never been higher. Listen to it on Spotify!
I would love to hear your thoughts, reply directly to the newsletter.
✅ Innovation Governance Program Level 1
I'm pleased to share that I've completed Level 1 of the Innovation Governance Program (iGP) from the Council of Canadian Innovators | Conseil canadien des innovateurs (CCI).
The program focuses on the realities of governing innovation-driven companies, for example; how boards support founders, oversee risk in fast-moving technology environments, and provide strategic guidance while maintaining effective oversight. It was valuable to learn alongside a strong cohort of operators and leaders from across Canada's innovation ecosystem.
Governance is an increasing area of focus for me and I'm particularly interested in the role boards can play in shaping the responsible development and deployment of AI technologies, ensuring innovation is paired with strong oversight, safety, and long-term strategic thinking.
I'm open to board or advisory board opportunities, particularly with companies building in AI, software, and emerging technologies. If you're building a board or governance structure and think my background could be helpful, I'd welcome the conversation.
🤖 AI Reads
Alibaba’s Qwen3.5-Medium Models Offer Sonnet 4.5 Performance
Notes: I was testing qwen3.5-35b-a3b this weekend. Full GPU offload with a decent context size is possible on a RTX 4090. It’s shockingly good. It may become my default coding model. The 35B-A3B model seems to be close enough to Claude Opus 4.6 per this study. Simon Willison shared news of potential personnel changes at Qwen. Unsloth posted an article on How to Run Qwen3.5 Locally. Find the models on HuggingFace. Read this X Post from Alibaba.
karpathy / autoresearch: AI Agents Running Research on Single-GPU NanoChat Training Automatically
Notes: Just goodness from Andrej Karpathy.
Introduction to LLM Post-Training Techniques
Notes: From Maxime Labonne at Liquid AI. Great ML scientist to follow.
Olmo WebGPU: Run Olmo Hybrid Directly in Your Browser
Notes: Always nice to see Edge AI projects being released by big research labs like AI2.
ruvnet / ruvector: RuVector is a Self-Learning, Vector Graph Neural Network, and Database
Notes: Interesting open source project.
builderz-labs / mission-control: Open-Source Dashboard for AI Agent Orchestration
Notes: I haven’t tried it yet because I already built my own control plane for my AI agents.
🔀 Other Reads
A Wave of Unexplained Bot Traffic Is Sweeping the Web
Notes: Hey look Ma, I made it! The vast majority of the traffic to my website and Substack over the last 90 days is from Lanzhou.
the peculiar case of japanese web design
Notes: An old article (2022), but a super fun read.
Signing off from Redmond, WA.