you'll be unstoppable.

KL night view! 🇲🇾🫶

Bullish for $TAI way to go! @tarsprotocol We cookin’

Day 2 of #MYB2025 The view is so amazing here in Exchange 106!

#MYB2025 hosting the institutional night here🫶 Been trying the local foods so far the Nasi Lemak is my fav!

Stoked to be at Malaysia Blockchain Week 2025! @MalaysiaBCW Always great to meet builders, exchange ideas, and vibe with the local scene 🇲🇾⚡️ At TARS AI @tarsprotocol, we’re not just building tools we’re shaping how people interact with AI. From agent marketplaces to powerful integrations, we’re on a mission to make AI more accessible, useful, and fun. We build fast, ship faster let’s link if you’re around! 🔥

Here at #MYBW2025 proudly representing @tarsprotocol sharing what we’ve been building at the intersection of Web3 and tech. Great to connect with curious minds, passionate builders, and new collaborators across the space. If you’re around, let’s connect!

This is my lecture from 2 months ago at @Cornell “How do I increase my output?” One natural answer is "I will just work a few more hours." Working longer can help, but eventually you hit a physical limit. A better question is, “How do I increase my output without increasing input as much?” That is leverage. We hear “leverage” so often that its implication is easy to overlook. My personal favorite categorization of leverage is by Naval Ravikant: human labor, capital, and code / media. Each has powered major waves of wealth creation in history. However, once a leverage source becomes popular (think YouTube channels today versus ten years ago) competition compresses the margin. So when a new leverage appears, it’s a rare chance for outsized gains. In this talk, I describe AI as that emerging leverage. An AI agent blends labor leverage (it does work for you and is permissionless) with code leverage (you can copy-and-paste it). It’s cliché to say AI will create massive wealth. But using this leverage lens lets us interpret the noisy AI news cycle in a consistent way and spot the real opportunities. Thanks @unsojo for hosting me!

Unfortunately, I won't be able to attend #ICML2025 in person due to visa delays. But I'm excited to share our paper in ICML: "Large Language Models are Demonstration Pre-Selectors for Themselves"! 🧠📄 💡 What if LLMs could help pick better examples for themselves? We propose FEEDER, a framework that lets LLMs pre-select a small but powerful subset of training data based on sufficiency and necessity, cutting costs, boosting performance, and generalizing across tasks. Come check it out!

Reinforcement Learning of Large Language Models, Spring 2025(UCLA) Great set of new lectures on reinforcement learning of LLMs. Covers a wide range of topics related to RLxLLMs such as basics/foundations, test-time compute, RLHF, and RL with verifiable rewards(RLVR).