The wiki

Modal’s co-founder and CEO works on making cloud computation easier to develop and scale. His earlier contributions include Spotify recommendation infrastructure, the similarity-search library Annoy, and the batch workflow engine Luigi.

Engineer and former robotics CTO whose Anthropic work combined minimal agent scaffolding with carefully designed tools, and whose public views emphasize reliability and auditable results.

Learning evolution’s constraints from protein sequences opens a route to structure prediction, molecular design, and discovery. Extending that approach to cells will require new data and experimental feedback.

AI engineer whose work spans NVIDIA Cosmos, Grok Imagine and mixture-of-experts language models, with a focus on interactive video and the language systems behind it.

Featherless AI cofounder and CEO, RWKV research collaborator, and advocate for affordable, multilingual open-model inference.

Applied AI scientist and writer working on reliable LLM systems, evaluation, and recommendations.

Evals turn questions about AI quality into repeatable tests: what a system can do, where it fails, and whether a change makes it better.

San Francisco Compute cofounder focused on making GPU capacity tradable and reducing the risks of long compute commitments.

Genesis co-founder and molecular machine-learning researcher whose work spans learned molecular representations and the accuracy needed for AI-assisted drug discovery.

Enterprise agents need useful context, bounded access, a workspace, and human oversight. Making those pieces fit an organization is as much a workflow problem as a model problem.

Formerly Metaphor, Exa builds neural web search and retrieval infrastructure for AI applications, alongside high-compute search and research agents.

A neural web-search company that evolved from a search engine for people into retrieval infrastructure for AI applications.

Understanding a query is only the start. Exa’s search ambitions depend on retrieving the right documents, clarifying intent and deciding where more computation is worth the cost.

An enterprise AI company whose Droids evolved from specialized software-development workflows into agents and coordinated software factories.

Moonlake cofounder and AI researcher working on interactive worlds, embodied-agent training environments, and structured representations for reasoning.

A deep-learning research and teaching lab founded by Jeremy Howard and Rachel Thomas, connecting free courses with open-source software and methods for working with less data and compute.

The Stanford computer scientist and World Labs cofounder helped build ImageNet and its shared recognition benchmark. Her work also spans scene description, dataset fairness, and spatial intelligence.

Anthropic engineer and Electron co-maintainer whose work brings AI task execution to the desktop.

A browser-based design and product development company expanding into prompt-to-app creation, design context for coding agents, and AI editing on its shared canvas.

An AI infrastructure company that expanded from open-model inference into workload optimization, compound AI systems and managed training.

Fireworks combines model serving, workload-specific optimization, and systems that connect models to tools and data. Its competitive bet is that inference engineering will matter more as open models improve.

FlashAttention preserves the attention computation while reducing memory traffic. Its development shows why mathematical ideas, hardware constraints and usable software have to advance together.

Lindy founder whose work on AI agents combines constrained workflows, model capabilities and software integrations.

OpenAI engineer whose work on Codex and Codex CLI informs his approach to sandboxing, network restrictions and human control of code-executing agents.

AI educator at UC Berkeley who advocates combining language models with ontologies to constrain agent behavior.

Astro creator and Flue developer bringing composable hooks, dynamic capabilities and a built-in harness to AI agents.

Part 2 moves from practical API integration to launch-day o1 questions: model stability, retrieval, batch processing, speech, and the cost and visibility of reasoning.

Part 3 of 4 follows builders from budget approvals and browser purchases to codebase search, guardrails, and open models—with usable context and reliable evaluation as recurring tests.

GPU marketplaces, image search, model training, and browser agents in Latent Space’s May 2023 conversation with Sharif Shameem.

AI’s accumulated breakthroughs, the Unix architecture behind personal agents, and the institutional work between technical capability and widespread adoption.