All Episodes
The Gist Talk — 309 episodes · page 1 of 4
The Weapon of Communication: Patrick Winston on Speaking effectively
Software Defined Chips
Coarse-Grained Reconfigurable Architecture: Taxonomy, Challenges, and Applications
Efficient Processing of Deep Neural Networks: A Comprehensive Survey
When Everyone Knows That Everyone Knows...
The Most Important Thing Illuminated: Uncommon Sense for Investors
LLM and World Models: Convergence, Divergence, and AGI Paths
LLM Inference Compiler Panorama: Research and Engineering Evolution
The AI-Native Fabless Chip Startup Blueprint
Groq Architecture Deep Dive and NVIDIA Acquisition Analysis
Huawei CloudMatrix 384 and Ascend 910C Architecture Analysis
The 95 Billion Dollar Dinner Plate Chip: Cerebras' Wafer-Scale AI Computing Architecture and Inference Performance Analysis
d-Matrix Corsair: An SRAM-Centric Digital In-Memory Compute Architecture
Tenstorrent AI Inference Architecture: Deep Dive into Tensix Dataflow
AI and the Economics of Production and Consumption Breakdown
The Evolution and Scaling of Google’s TPU Supercomputers
The Mathematics of LLM Training and Inference
The Foundation of an AI-Native Company: Closed Loops and Intelligence Layers
The Shape of the Company as the AI Moat: The Next Biggest Moat in AI
The Race to the Bottom: Risk and Laxity in Finance
The Architecture of Innovation and the Mechanics of Bubbles
vLLM Plugin System and Hardware Pluggability Architecture
Howard Marks: AI Hurtles Ahead, The Evolution of Autonomous Intelligence
Howard Marks: The Evolution and Cycles of Private Credit
Howard Marks: Is It a Bubble? The Nature of AI Euphoria
Howard Marks: Cockroaches in the Coal Mine
The 2026 Stanford AI Index: Trends, Investment, and Global Impact
Agentic Reasoning for Large Language Models
Dynamic Hedging: Part 4
Dynamic Hedging: Part 3
Dynamic Hedging: Part 2
Dynamic Hedging Part 1: Foundations of Market Reality and Risk
Volatility Smile and Delta Hedging: Intimate with the Vol Surface
Howard Marks: You Can't Predict. You Can Prepare.
Howard Marks: The Limits to Negativism
Howard Marks: The Paradox of Liquidity
Howard Marks: There They Go Again, on Market Cycles
Howard Marks: On the Couch
Howard Marks: Sea Change, The New Era of Investment Strategy
Howard Marks: Calibrating, Striking the Balance Between Offense and Defense
Howard Marks: The Asymmetry of Risk and the Unknowable Future
Howard Marks: Taking the Temperature, Mastering Market Cycles and Investor Psychology
Howard Marks: What Really Matters, Long-Term Thinking in a Short-Term World
Howard Marks: Fewer Losers More Winner
AI Traffic Patterns and AI Switch Design Implications
AI network switches
DeepSeek-V4: Efficient Million-Token Context Intelligence
Claude Code Source Code and Architecture Analysis
Dimon’s $1.5 Trillion Gambit: Can Private Sector Resiliency Save the West from "Vassal State" Decline?
From Entropy to Epiplexity: Rethinking Information for Computationally Bounded Intelligence
Challenges and Research Directions for LLM Inference Hardware
DeepSeek Engram: Conditional Memory via Scalable Lookup
End-to-End Test-Time Training for Long Context
Computational intelligence in data-driven
Notes on Complexity
Complexity and the Econominy
A Tail Hedging Strategy
Trading volatility spread
option pricing, volatility, and advanced trading strategies
Dynamic Hedging
vLLM - LLM Serving Optimization: Paging, Routing, and Ranking
Jamba-1.5: Hybrid Transformer-Mamba Models at Scale
Google's Titans+Miras: Learning to Memorize at Test Time
LLM Architectures: Attention, Mamba, and Efficiency Tradeoffs
Grouped-Query Attention: Speed and Quality Through Uptraining
Cross-Layer Attention for KV Cache Optimization
Performers: Linear Transformers with Orthogonal Random Features
Linear Attention Transforms RNNs and Accelerates Autoregression
A Comprehensive Survey of Efficient Transformer Models
DeepSeek-V3: A Strong and Efficient MoE Language Model
Cake: Computation and I/O Aware KV Cache Loader
vAttention: Dynamic LLM Memory Without PagedAttention
Attention Is All You Need: The Transformer
Multi-Token Prediction for Efficient LLM Inference
Long Short-Term Memory and Recurrent Networks
The Theory of Poker: Deception and Expectation
A Definition of AGI
The Ultra-Scale Playbook Training LLMs on GPU Clusters
AIBrix: Scalable Control Plane for vLLM
Offloading LLM Attention: Q-Shipping and KV-Side Compute
PagedAttention: Efficient LLM Memory Management - Part 2
PagedAttention: Efficient LLM Memory Management
DeepSeek Deployment with SGLang: Disaggregation and Expert Parallelism
Markov Chains (and HMM) for Quantitative Finance Modeling
Cybernetic Analysis for Stocks and Futures - Part 2
Cybernetic Analysis for Stocks and Futures - Part 1
Quant 101
Trading Option Implied Volatility
Poker and Trading: Mastering Optimal Action in Uncertainty
Trend Following - Part 3: Managed Futures Strategies
Trend Following - Part 2: Pioneers and Systematic Trading
Trend Following - Part 1: The Philosophy and Practice
Stock Prediction Using Raw OHLCV and Triple Barrier Labeling
Consequences of Fat Tails
The Laws of Trading: Part 2 - Models, Markets, and Trading Systems
The Laws of Trading: Part 1 - Edge, Risk, and Liquidity
How OpenAI Uses Codex
Smart Portfolios: Part 4 - The Theory of Portfolio Rebalancing and Substitution
Smart Portfolios: Part 3 - Forecasting Returns, Models, and Active Management
Smart Portfolios: Part 2 - Top-Down Approach and Asset Allocation