The wiki

Neural-network researcher, OpenAI cofounder and SSI leader whose work spans image recognition, sequence learning and the problem of reliable generalization.

An AI research company whose work spans reasoning models, coding-agent interfaces, and open tools for automated research.

From image search to conversational characters, useful AI depends on matching hardware, model size and training data to the task—and meeting its response-time budget.

Part 2 of 2 follows inference engineering from distributing a model across GPUs to keeping video coherent, training for efficient serving, and preserving knowledge across requests.

Part 1 follows a long request through caching, speculative decoding and production serving, then examines quantization fidelity, meaningful speed comparisons and the constraints of local inference.

Poolside Infrastructure Company, called “PIC infraco” in Latent.Space, develops Project Horizon, an AI data center campus in West Texas.

How Brex combines a small AI team with shared infrastructure—and handles the harder work of evaluation, institutional knowledge, and code ownership.

How Gemini turns a question into an adaptive investigation—and the model, interface, evaluation, and execution choices that make it useful.

Streaming data, reproducible experiments and persistent agents serve a common goal: shorten the path from a research idea to a trustworthy model improvement.

Jason Liu’s open-source structured-output SDK turns language-model responses into typed data, with Pydantic validation and retries.

Jason Liu’s April 2024 conversation on turning model outputs into usable data, keeping libraries focused, and learning from experiments.

Roboflow computer-vision contributor and RF-DETR coauthor whose work includes real-time detection, deployment fixes, and technical explanations of segmentation and keypoint uncertainty.

NVIDIA engineer working on distributed inference and the serving infrastructure that lets coding agents reuse context, preserve tool interactions, and run across open-model backends.

Co-founder of Qodo, formerly CodiumAI, and co-author of AlphaCodium, whose work centers on testing, code review, and the engineering needed to make coding agents reliable.

Daytona co-founder who led its pivot from developer environments to AI-agent sandboxes, drawing on earlier work at Codeanywhere, Shift and Infobip.

AI engineer and Instructor contributor whose work spans structured outputs, retrieval evaluation and general-purpose agents.

AI safety researcher and builder of the Factorio Learning Environment, studying open-ended agent capabilities and the limits of model auditing.

AI researcher studying what language models and embeddings retain, how that information can be recovered, and how data shapes model capabilities.

Redpoint investor and Unsupervised Learning host examining how AI capabilities become useful products, particularly in enterprise software and healthcare.

Intuitive prompt exploration, contested evaluations, open research, and the risks of agents that see only part of a task.

Railway’s founder combines hands-on infrastructure work with a push toward agent-driven development, while insisting that automated changes need safe testing and rollout mechanisms.

Engineering leader at Elicit whose public work explains how to build reliable language-model applications and hire engineers for them.

Brex’s CTO in January 2026, whose AI work spans internal infrastructure, operational automation and specialist-agent systems.

Designer and developer building practical AI tools, with work spanning invoicing, model integrations and scientific communities.

Abridge’s vice president of product on ambient healthcare AI, clinical evaluation, and why complex products still need written clarity.

Developer-tool founder and AI product leader whose work includes v0, AI SDK and Turborepo.

Computer-vision researcher and Voxel51 co-founder whose joint work measures classification-dataset difficulty and screens for potential label mistakes.

Creator of Instructor, the Python library for structured language-model outputs, known for consulting and teaching on retrieval, evaluation, and measurable product outcomes.

Dyna Robotics co-founder and chief scientist applying large-scale video learning to physical robot control.

Author of The Illustrated Transformer, developer of Ecco, and coauthor of Hands-On Large Language Models, Jay Alammar explains how language models work and how developers can apply them.