Rami's Readings #89 - Announcing Apéro & Intellect #10 in NYC!
Announcing Apéro & Intellect in New York City and the latest on AI, LLMs, and more.
Welcome to Rami’s Readings #89 - 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.
Apéro & Intellect Is Coming to New York City!
Apéro & Intellect is a curated series of intimate gatherings discussing Artificial Intelligence designed to stimulate both your mind and your palate. I am pleased to announce Apéro & Intellect #10 will take place in 🗽 New York City on Tuesday, December 10th, at 8am. It will be fun!
🤖 AI Reads
The Way We Interact With AI Chatbots Is Changing
Notes: A short summary of the latest on AI, LLMs, and Agents. If you’re new to the newsletter, this read is easily digestible.
Scientists Establish the Best Algorithm for Traversing a Map
Notes: This article is not about AI, but the principles are relevant to those buildings AI systems. A tip for those interviewing to join my engineering team: most of my interview questions test whether you can recognize that a different data structure would solve an algorithmic problem more efficiently.
Qwen2.5-Coder Series: Powerful, Diverse, Practical.
Notes: Qwen2.5-Coder is turning out to be an exceptionally good open source coding model. It feels better than even Claude Sonnet 3.5.
OmniVision-968M: World's Smallest Vision Language Model
Notes: Super small visual language model. You can run this mobile devices. Further evidence for #Edge AI.
LLaMA-Mesh: Unifying 3D Mesh Generation with Language Models
Notes: From the Nvidia Toronto Research group. This paper surprised me and it is provoking a number of ideas. My first coding project as a teenager was building a game engine.
Our work is the first to demonstrate that LLMs can be fine-tuned to acquire complex spatial knowledge for 3D mesh generation in a text-based format, effectively unifying the 3D and text modalities.
Athene-V2: Advancing Beyond the Limits of Scaling with Post-training
Notes: This updated model beats GPT4o and Llama 3.1 405B Chat on a number of benchmarks.
💼 Business Reads
Expat Taxes Are Too Complicated
Notes: From first-hand experience, expat taxation is unnecessarily complicated. As a Canadian expat in the US, I paid more money to the accounting firm to correctly file everything than the additional taxes levied. I shudder at the thought of having to go through the process again for another move. Finding all the data, filling out all the forms, so much energy expended. My own experience is why I partially agree with Tyler: making this process simpler would facilitate US investments around the world.
Founder / Advisor Standard Template (FAST)
Notes: For my readers in a long-term advisory relationship with a startup, Founder Institute has a standard template you can use to formalize your relationship.
🎨 Culture Reads
Casey’s video strongly resonated with me. Always worth a watch! ❤️
Signing off from somewhere over Arizona.