| 08-30-25 | Using Claude Code from Anywhere |
| 12-03-23 | ChatGPT After One Year |
| 12-01-23 | Data Quality in LLMs |
| 11-30-23 | Discord and AI GTM |
| 11-24-23 | How AI Changes Workflows |
| 11-22-23 | Strategies for the GPU-Poor |
| 11-18-23 | The Model is Not the Product |
| 11-17-23 | The AI-Neid |
| 11-16-23 | Model Merge - (Frankenmerge) |
| 11-15-23 | The Cost of Index Everything |
| 11-13-23 | Copilot is an Incumbent Business Model |
| 11-09-23 | AI Agents Today |
| 11-08-23 | Norvig's Agent Definition |
| 11-07-23 | The Context Length Observation |
| 11-05-23 | Improving RAG: Strategies |
| 11-03-23 | Lessons from llama.cpp |
| 11-02-23 | Why Model Evaluation is Difficult |
| 11-01-23 | Mechanical Turks |
| 10-30-23 | What If OpenAI Builds This? |
| 10-28-23 | Infrastructure as Code Will be Written by AI |
| 10-26-23 | Between Images and Text, CLIP |
| 10-24-23 | Tech Invariants |
| 10-23-23 | Horizontal Tuning: Instruction, Chat, and What Else? |
| 10-22-23 | Retrieval Augmented Generation |
| 10-19-23 | Benefits of Small LLMs |
| 10-18-23 | Can OpenAI Win Consumer and Enterprise? |
| 10-16-23 | Revision: Generative text-to-UI |
| 10-14-23 | An Intelligent Wikipedia |
| 10-13-23 | The Half-Life of the AI Stack |
| 10-09-23 | Moravec's Paradox |
| 10-02-23 | Generative Interfaces |
| 09-30-23 | Compression / Learning Duality |
| 09-29-23 | Is AI a Platform Shift? |
| 09-28-23 | Passkeys, Crypto, and Signing AI Content |
| 09-27-23 | Is Data Still a Moat? |
| 09-26-23 | Multi-Modal AI is a UX Problem |
| 09-20-23 | AI Biographers |
| 09-19-23 | Customized End User Software (with AI) |
| 09-16-23 | The Age-old Resistance to Generated Code |
| 09-14-23 | Undetectable AI |
| 09-11-23 | Fine-tuning Stable Diffusion XL with Personal Photos |
| 09-09-23 | Beyond Prompt Engineering |
| 09-05-23 | Type Constraints for LLM Output |
| 09-01-23 | Capital Intense AI Bets |
| 08-30-23 | Llama 2 in the Browser |
| 08-29-23 | The Contrarian Strategy of OpenAI |
| 08-27-23 | AI and Text-First Interfaces |
| 08-25-23 | The Free Lunch of Model Distillation |
| 08-21-23 | A Model API Gateway for 20+ LLMs |
| 08-16-23 | What is a Prompt Engineer? |
| 08-13-23 | My Everyday LLM Uses |
| 08-11-23 | Llama/Unix |
| 08-10-23 | Deterministic, Structured LLM Output |
| 08-08-23 | A Fine-Tuning Marketplace |
| 08-01-23 | Automatic and Universal Adversarial Prompts |
| 07-20-23 | Robots.txt for LLMs |
| 07-19-23 | Why Did Meta Open-Source Llama 2? |
| 07-15-23 | Scale to Zero for AI Workloads |
| 07-07-23 | The Anti-AI Movement |
| 07-04-23 | Where AI Fits in Engineering Organizations |
| 06-30-23 | Personal Lessons From LLMs |
| 06-29-23 | Overcoming LLM Hallucinations |
| 06-25-23 | Model Evaluation is (Still) An Art |
| 06-22-23 | No Feature is Free (Especially AI Ones) |
| 06-21-23 | Mixture of Experts: Is GPT-4 Just Eight Smaller Models? |
| 06-20-23 | The LLaMA Ecosystem |
| 06-17-23 | The Low-Background Steel of AI |
| 06-16-23 | Why Does Every AI Cycle Start With Chat? |
| 06-15-23 | A Token Efficient Language for LLMs |
| 06-12-23 | It’s Too Early To Call Winners in AI |
| 06-09-23 | The Myth of the AI Infrastructure Phase |
| 06-08-23 | LLMs For Software Portability |
| 06-07-23 | ChatGPT Plugins Don't Have PMF |
| 06-06-23 | Levels of Autonomy in AI Agents |
| 06-05-23 | The Problem with Tokenization in LLMs |
| 06-04-23 | What Diffusion Models Can Teach Us About LLMs |
| 06-03-23 | Sequence and Version Control Models |
| 05-31-23 | Faster Horses: AI Products That Companies Think They Want |
| 05-29-23 | AI Means More Developers |
| 05-27-23 | Prompt Engineering is Configuration Engineering |
| 05-26-23 | SEO Inside AI |
| 05-24-23 | The ChatGPT Plugin Specification |
| 05-21-23 | The New Tax on Engineering Against the Grain |
| 05-19-23 | On Regulating AI |
| 05-18-23 | On Device AI? |
| 05-17-23 | A List of Leaked System Prompts |
| 05-16-23 | Intercloud Brokers |
| 05-15-23 | React LLM: Run Models in the Browser with Headless Components |
| 05-14-23 | Context-Free Grammar Parsing with LLMs |
| 05-12-23 | StackOverflow/ChatGPT |
| 05-11-23 | Self-hosted Compilers and Bootstrapped AI |
| 05-10-23 | Unix Philosophy for AI |
| 05-08-23 | The New AI Moats |
| 05-05-23 | ReLLM: Exact Structure for Large Language Model Completions |
| 05-04-23 | llm.ts |
| 04-30-23 | Implementing LLMs in the Browser |
| 04-26-23 | Probabilistic Data Structures and LLMs |
| 04-21-23 | Autonomous LLM Agents Are At Least 10 Years Out |
| 04-19-23 | Sandbox Your Prompts |
| 04-17-23 | Jevons Paradox and LLMs |
| 04-14-23 | Synthetic Data From Compilers |
| 04-13-23 | Foundational Models Are Not Enough |
| 04-10-23 | A List of 1 Billion+ Parameter LLMs |
| 04-09-23 | No GPUs before Product-Market Fit |
| 04-07-23 | Buyers in the Foundational Model Stack |
| 04-05-23 | A High-level LLMOps Architecture |
| 04-03-23 | The Automation Frontier |
| 04-02-23 | Why Open-Source a Model? |
| 03-30-23 | The AI Partnership Race |
| 03-29-23 | A Hacker's Guide to LLM Optimization |
| 03-27-23 | Code, not Chat, in Generative AI |
| 03-26-23 | Distributed Systems and AI |
| 03-23-23 | Are Incumbents Accruing All The AI Value? |
| 03-22-23 | Model Arbitrage |
| 03-21-23 | Modeling Context Length vs. Information Retrieval Cost in LLMs |
| 03-17-23 | Foundational Models Are Commodities |
| 03-16-23 | On OpenAI's Kubernetes Cluster |
| 03-15-23 | Choosing the Right Model |
| 03-13-23 | On Prompt Injection |
| 03-12-23 | Local LLaMA on a Mac M1 |
| 03-11-23 | Automatic1111 and AI Aggregators |
| 03-09-23 | Chain of Thought Paradigms in LLMs |
| 03-05-23 | Will The AI Stack Be Open Or Closed? |
| 03-02-23 | ChatML and the ChatGPT API |
| 02-25-23 | Commoditization of Large Language Models: Part 3 |
| 02-19-23 | Why Python Won't Be the Language of LLMs |
| 02-16-23 | Why ChatGPT Needs AuthZ |
| 02-12-23 | LLM Ops, Part 1 |
| 02-10-23 | Multi-Model vs. Multi-Cloud |
| 02-05-23 | Composable Models |
| 01-28-23 | Overview of GPT-as-a-Backend |
| 01-23-23 | Prompt Engineering Shouldn't Exist |
| 01-22-23 | GPT Lineage |
| 01-14-23 | Garbage In, Garbage out? |
| 01-12-23 | Minix and nanoGPT |
| 01-10-23 | Lessons from the Last AI Cycle |
| 01-08-23 | Fine-Tuning an OCR Model |
| 01-02-23 | A New ML Stack |
| 12-30-22 | Local AI: Part 2 |
| 12-29-22 | Local AI: Part 1 |
| 12-26-22 | Turing Social: Twitter, For Bots |
| 12-21-22 | ML Developer Experience |
| 12-15-22 | AI-driven Interfaces |
| 12-14-22 | LAION, The Pile, and more datasets |
| 12-12-22 | Lessons from Lensa |
| 12-08-22 | Spam Filtering AI Content |
| 12-06-22 | Stack Overflow Bans ChatGPT |
| 12-05-22 | Will LLMs Disrupt Google Search? |
| 12-03-22 | A Conversation with ChatGPT |
| 11-27-22 | Human-in-the-Loop and Other AI Mistakes |
| 11-19-22 | Generative AI Value Chain |
| 11-13-22 | LLMs for Code |
| 10-21-22 | AI Will Write Most Code |
| 10-19-22 | AI Scaling Laws |
| 09-12-22 | TensorFlow vs. PyTorch |
| 07-25-22 | Defensible Machine Learning Model Naming |
| 06-22-22 | How to Use GitHub Copilot Effectively |
| 11-09-21 | Open-sourced GPT-J |
| 08-17-21 | One Month of Using GitHub Copilot |
| 07-10-21 | GitHub Copilot |
| 06-26-21 | Machine Learning Operations (MLOps), Convergent or Divergent? |
| 03-14-21 | ScapeNet: Real-time object detection in RuneScape |