Alec Radford
5 sourcesGenerative pretraining, multimodal models
Important because several of the modern foundation-model playbooks trace back to work he helped drive, especially around generative pretraining and multimodal transfer.
AI Research Map
The goto map for finding the people shaping AI.
500AI is a curated guide to the researchers, labs, topics, and source materials that matter most. It is built for people who want to move from vague awareness to a clear understanding of who matters, why they matter, and where to start.
At a glance
Researchers
3,615
Topics
12
Labs & ecosystems
13
Source clusters
156
How to use 500AI
Start with a topic if you know the area but not the people.
Start with a lab if you want to understand a cluster of work.
Use researcher pages to jump into canonical papers and adjacent people.
Researchers To Start With
A deliberately balanced starting set spanning frontier labs, open communities, evaluation, alignment, systems, and multimodal work.
Dario Amodei
3 sourcesAlignment, post-training, frontier LLMs
A high-signal figure for understanding the frontier model era because his work sits at the intersection of scaling, post-training, and deployment-risk framing.
Start HereAnthropic company
Stella Biderman
5 sourcesOpen-source LLMs, datasets
A key open-model ecosystem builder whose work matters because it combines research, public infrastructure, and field-level coordination rather than isolated paper output alone.
Andrej Karpathy
3 sourcesDeep learning engineering, LLM education
Important not only for his direct research contributions, but for translating frontier deep-learning ideas into builder intuition that spreads across the industry.
Start HereAndrej Karpathy
Pushmeet Kohli
4 sourcesRobotics, vision, structured prediction
A strong person to follow if you want to understand how frontier AI gets pushed into science, security, and trustworthy deployment rather than staying inside benchmark culture.
Chris Olah
4 sourcesMechanistic interpretability, visualization
One of the clearest interpreters of neural-network internals, especially in the line of work that turned interpretability into a concrete research agenda rather than a vague aspiration.
Start HereFeature Visualization
Jeff Dean
4 sourcesML systems, large-scale infrastructure
Foundational less for any single public paper than for shaping the infrastructure, engineering culture, and systems thinking that make frontier-model research possible.
Arthur Mensch
4 sourcesOpen-weight LLMs
One of the clearest people to track if you want to understand how frontier open-weight labs balance model quality, deployment speed, and product ambition.
Start HereMistral AI
Kaiming He
3 sourcesComputer vision, representation learning
A foundational computer-vision researcher whose work on representations and architectures still shapes modern pretraining and perception systems.
Start HereKaiming He
Topics
The fastest way to understand the landscape by research area.
Open Models
1,545 researchers
Researchers pushing open-weight language, code, and multimodal systems that the broader ecosystem can inspect and build on.
Multimodal
1,373 researchers
People building systems that connect language with images, audio, video, and embodied perception.
Systems & Infrastructure
417 researchers
Researchers who make large-scale training and inference practical through architecture, kernels, sharding, and serving work.
Evaluation & Benchmarks
244 researchers
People building the measurement systems, benchmarks, and red-team style checks used to understand AI systems.
Post-Training & Alignment
148 researchers
Researchers shaping model behavior after pretraining, from instruction tuning and preference learning to scalable oversight.
Code Models
139 researchers
Researchers behind code-specialized models, datasets, and evaluation setups for software engineering tasks.
Agents & Reasoning
110 researchers
People exploring planning, tool use, and reasoning-heavy model behavior for longer-horizon tasks.
Reinforcement Learning
90 researchers
Researchers working on decision-making, planning, self-play, and RL methods that still shape modern AI systems.
Labs & Ecosystems
Research clusters, institutions, and model ecosystems worth tracking.
Google DeepMind
1,185 researchers
Researchers shaping frontier multimodal, RL, and scientific AI systems across the DeepMind lineage.
Meta
583 researchers
Builders behind Llama, Segment Anything, and large-scale open-weight model research.
447 researchers
People driving Google’s open-model, multimodal, and large-scale infrastructure work outside the DeepMind label.
OpenAI
289 researchers
Researchers behind GPT-era training, post-training, multimodal systems, and model evaluation work.
DeepSeek
195 researchers
Teams pushing frontier open-model reports, efficient training, and high-capability open-weight systems.
Qwen
79 researchers
Researchers working on the Qwen family of open-weight language and multimodal systems.
Anthropic
71 researchers
Alignment, post-training, and frontier assistant researchers with a strong safety and behavior focus.
AI21
61 researchers
Researchers exploring long-context hybrid architectures and practical language model deployment.
Canonical Starting Points
The papers and project pages most frequently used to anchor the researcher graph.
Gemini: A Family of Highly Capable Multimodal Models
1128Referenced across 1,128 researcher pages
The Llama 3 Herd of Models
482Referenced across 482 researcher pages
Llama (site)
482Referenced across 482 researcher pages
Gemma (docs)
359Referenced across 359 researcher pages
GPT-4 Technical Report
204Referenced across 204 researcher pages
DeepSeek-V3 Technical Report
195Referenced across 195 researcher pages
DeepSeek (project)
195Referenced across 195 researcher pages
Gemma 2: Improving Open Language Models at a Practical Size
141Referenced across 141 researcher pages