The wiki

LLaVA coauthor and multimodal AI researcher whose work spans visual instruction tuning, vision-language evaluation and Grok model development.

Part 2 follows coding agents from demonstrating a change to working inside an enterprise—and asks which harness investments will remain useful as models improve.

Ryan Lopopolo’s experiment began with a constraint: write none of the code himself. Part 1 follows the resulting work on builds, observability, review, and Symphony’s agent orchestration—and the human attention those systems still require.

Creator of LangChain and co-founder of the company, Chase develops software for connecting language models to data, controlling agent execution and managing longer tasks.

Developer Experience leader at Together AI who builds open-source AI applications and teaches developers how to use models in working products.

MIT computational chemist combining quantum chemistry, machine learning, and experimental collaboration to discover molecules and materials.

Builder of Smithery and co-founder of Jenni, with earlier research in generative neural networks and efficient language models.

A roundtable on local tools, fast decisions, human review and the permissions that shape everyday assistants.

Elicit’s approach combines software engineering, curiosity about model capabilities, and product judgment about when to recover, experiment, or report a failure.

From enterprise search to coding agents, durable advantages depend on integration work, product habits, research progress, and the judgment needed to use the results.

Pokémon exposes the gap between choosing a goal and reliably acting on it—and makes improvements in long-running agents visible.

Ari Weinstein and Nikunj Handa explain the interface and API choices behind faster agents, and where their limits remain.

Jason Ma explains human-video pretraining, robot demonstrations, and the hardware and workflow decisions behind Dyna’s physical agents.

From asking questions of a textbook to generating two-person audio discussions, NotebookLM’s development joined model capability with editorial judgment—and a deliberately simple interface.

Inside WorldPrompt, real-time diffusion, robot simulation, and the limits of predicting what happens next.

An AI platform for sharing models, datasets and applications, with research spanning language, multimodality and robotics.

Researcher whose collaborative work on Idefics, WebSight and open training corpora examines how data, architecture and evaluation shape vision-language models.

Humanloop developed tools for prompt management, LLM evaluation and production observability before announcing its team's move to Anthropic in 2025.

From selecting wind-tunnel experiments to evaluating generative applications, Humanloop’s story follows a changing bottleneck: turning model behavior into evidence a team can act on.

Keycard cofounder focused on task-scoped access, accountable delegation, and control over autonomous AI agents.

Leash Bio co-founder building the experimental data and cross-disciplinary teams behind models of protein–molecule interactions.

Promptfoo co-founder whose work connects systematic LLM evaluation with application security and adversarial testing.

Part 1 follows the VAE retrospective, Würstchen’s compressed image generation, concept-based controls for diffusion, the compression argument for unsupervised learning, and the opening of Christian Szegedy’s adversarial-examples retrospective.

Part 1 follows the path from rebuilding an NLP course after ChatGPT to web agents, social evaluation, code optimization and the compute constraints shaping academic research.

Part 2 follows tiny image perturbations, tokens that provide working space, selective training data, and cheaper context extension—ending with the distinction between a larger window and the ability to use it.

Part 2 moves from the experiments academia can afford to run into SWE-Bench, contamination detection, and GAIA’s tests of everyday assistance.

Part 3 of 4 moves from the limits of longer context to synthetic-data selection, adaptive KV caches and distributed training, then follows Sasha Rush’s tutorial on recurrent models for language and diffusion.

Part 4 follows stepwise reasoning feedback into deployment safety, browser automation, and multi-agent software development, then closes with the reversal curse: why learning a fact in one direction may not teach its reverse.

Researcher working on reinforcement learning, cognitive maps, and the evaluation and design of planning systems built with language models.

Sequence-to-sequence researcher, founding OpenAI research director, and cofounder and CEO of Safe Superintelligence Inc.