# AKI AI Scientists, Engineers & Talent Intelligence Suite™ (Top 100)
> **Canonical Grounding:** https://aki1k.com/talent  
> **API Gateway:** https://api.aki1k.com/v1/talent  
> **Transfer Feed:** https://api.aki1k.com/v1/talent/transfers  
> **License & Attributions:** Creative Commons Attribution 4.0 International (CC BY 4.0)  
> **Audit Status:** Verified Telemetry • ZIP-1.0 Institutional Protocol  
> **Updated:** September 2026  

---

## Executive Summary & Index Architecture
The AKI AI Talent 100™ is the institutional benchmark quantifying human intellectual capital across global artificial intelligence laboratories. It evaluates top researchers, systems architects, and research leaders across three empirical vectors:
1. **Scientist Score (40%):** Novel algorithmic originality, fundamental theoretical breakthroughs (e.g., Transformers, FlashAttention, JEPA, Diffusion, PPO), and seminal paper citation velocity.
2. **Engineer Score (30%):** Low-level systems execution, distributed training infrastructure, custom CUDA kernels, memory-hierarchy efficiency, and framework creation.
3. **12-Month Momentum Velocity (30%):** Recent citation trajectory, active model deployment impact, lab migration seniority, and breakthrough velocity.

---

## Global AI Talent 100 Index

| Rank | Name | Primary Role | Current Affiliation | Composite | Scientist | Engineer | 12M Momentum | Key Architectural Contributions | Profile URI |
| :---: | :--- | :--- | :--- | :---: | :---: | :---: | :---: | :--- | :--- |
| **#1** | [Ilya Sutskever](https://aki1k.com/talent/ilya-sutskever) | `Research Leader` | Safe Superintelligence (SSI) | **99.6** | 99.8 | 98.2 | +320% | AlexNet, Sequence-to-Sequence (Seq2Seq), GPT-1 through GPT-4, Superalignment Theory | [API](https://api.aki1k.com/v1/talent/ilya-sutskever) |
| **#2** | [Yann LeCun](https://aki1k.com/talent/yann-lecun) | `Scientist` | Meta FAIR & NYU | **99.1** | 99.5 | 95.0 | +210% | Convolutional Neural Networks (LeNet), JEPA World Models, Self-Supervised Learning, Turing Award 2018 | [API](https://api.aki1k.com/v1/talent/yann-lecun) |
| **#3** | [Andrej Karpathy](https://aki1k.com/talent/andrej-karpathy) | `Research Leader` | Eureka Labs | **98.8** | 97.2 | 99.4 | +410% | CS231n Deep Learning Course, Tesla Full Self-Driving Vision, nanogpt & llm.c, Eureka Labs AI Native Education | [API](https://api.aki1k.com/v1/talent/andrej-karpathy) |
| **#4** | [Demis Hassabis](https://aki1k.com/talent/demis-hassabis) | `Research Leader` | Google DeepMind | **98.5** | 99.2 | 96.0 | +290% | AlphaGo & AlphaZero, AlphaFold 1-3 (Nobel Prize Chemistry 2024), Gemini Architecture, Deep Q-Networks (DQN) | [API](https://api.aki1k.com/v1/talent/demis-hassabis) |
| **#5** | [Noam Shazeer](https://aki1k.com/talent/noam-shazeer) | `Engineer` | Google DeepMind | **98.4** | 97.8 | 99.8 | +380% | Attention Is All You Need (Transformer), Mixture of Experts (MoE), SwiGLU & Multi-Query Attention, Character.AI Engine | [API](https://api.aki1k.com/v1/talent/noam-shazeer) |
| **#6** | [Tri Dao](https://aki1k.com/talent/tri-dao) | `Engineer` | Together AI & Princeton | **98.2** | 97.0 | 99.9 | +540% | FlashAttention 1, 2 & 3, Mamba State Space Models, CUDA Kernel Optimization, Monarch Matrices | [API](https://api.aki1k.com/v1/talent/tri-dao) |
| **#7** | [Dario Amodei](https://aki1k.com/talent/dario-amodei) | `Research Leader` | Anthropic | **97.9** | 98.0 | 96.5 | +310% | RLHF (Reinforcement Learning from Human Feedback), Claude Model Family, Constitutional AI, Deep Speech | [API](https://api.aki1k.com/v1/talent/dario-amodei) |
| **#8** | [Tim Dettmers](https://aki1k.com/talent/tim-dettmers) | `Engineer` | CMU & bitsandbytes | **97.4** | 95.8 | 99.2 | +460% | bitsandbytes LLM.int8(), QLoRA 4-bit Quantization, NF4 Data Type, 8-bit Optimizers | [API](https://api.aki1k.com/v1/talent/tim-dettmers) |
| **#9** | [Geoffrey Hinton](https://aki1k.com/talent/geoffrey-hinton) | `Scientist` | University of Toronto | **97.1** | 99.9 | 90.0 | +180% | Backpropagation Algorithm, Boltzmann Machines, Deep Belief Networks, Turing Award & Nobel Physics 2024 | [API](https://api.aki1k.com/v1/talent/geoffrey-hinton) |
| **#10** | [Sholto Douglas](https://aki1k.com/talent/sholto-douglas) | `Rising Star` | Google DeepMind | **96.8** | 94.5 | 98.6 | +580% | Scaling Laws Exploration, Infini-attention, Recurrent Transformer Variants, Frontier Compute Analysis | [API](https://api.aki1k.com/v1/talent/sholto-douglas) |
| **#11** | [Yoshua Bengio](https://aki1k.com/talent/yoshua-bengio) | `Scientist` | Mila & Université de Montréal | **96.5** | 99.4 | 91.0 | +200% | Neural Language Models, Generative Adversarial Networks (GANs), Word Embeddings, Turing Award 2018 | [API](https://api.aki1k.com/v1/talent/yoshua-bengio) |
| **#12** | [Fei-Fei Li](https://aki1k.com/talent/fei-fei-li) | `Research Leader` | World Labs & Stanford HAI | **96.2** | 98.2 | 94.0 | +350% | ImageNet Dataset & Benchmark, Spatial Intelligence 3D Models, World Labs Vision, Stanford Human-Centered AI | [API](https://api.aki1k.com/v1/talent/fei-fei-li) |
| **#13** | [Aravind Srinivas](https://aki1k.com/talent/aravind-srinivas) | `Research Leader` | Perplexity AI | **95.8** | 95.0 | 98.0 | +490% | Perplexity Answer Engine, Agentic Live Search RAG, Contrastive Learning, Transformer Vision Backbones | [API](https://api.aki1k.com/v1/talent/aravind-srinivas) |
| **#14** | [Arthur Mensch](https://aki1k.com/talent/arthur-mensch) | `Research Leader` | Mistral AI | **95.5** | 96.5 | 97.2 | +420% | Mistral 7B & Mistral Large, Mixtral 8x7B Sparse MoE, Flamingo Multimodal Architecture, Codestral | [API](https://api.aki1k.com/v1/talent/arthur-mensch) |
| **#15** | [Clément Delangue](https://aki1k.com/talent/clement-delangue) | `Research Leader` | Hugging Face | **95.1** | 92.0 | 98.5 | +340% | Global Open-Source AI Hub, Transformers Library Registry, Open LLM Leaderboard, Democratized Model Access | [API](https://api.aki1k.com/v1/talent/clement-delangue) |
| **#16** | [Guillaume Lample](https://aki1k.com/talent/guillaume-lample) | `Engineer` | Mistral AI | **95.0** | 95.8 | 98.9 | +390% | Mistral Architecture & Serving, XLM Cross-lingual Language Modeling, LLaMA 1 Lead Architect, Byte-Pair LLM Tokenization | [API](https://api.aki1k.com/v1/talent/guillaume-lample) |
| **#17** | [Lilian Weng](https://aki1k.com/talent/lilian-weng) | `Scientist` | Independent AI Safety Research | **94.8** | 96.8 | 94.5 | +280% | Lil'Log AI Technical Synthesis, LLM-Powered Autonomous Agents Architecture, OpenAI Safety Systems & Red-Teaming, Diffusion Models Foundations | [API](https://api.aki1k.com/v1/talent/lilian-weng) |
| **#18** | [Jim Fan](https://aki1k.com/talent/jim-fan) | `Physical AI Scientist` | NVIDIA GEAR Lab | **94.6** | 95.2 | 97.5 | +470% | NVIDIA Project GR0OT Humanoid Model, Voyager Lifelong Learning Agent, MineDojo Open Benchmark, Isaac Sim Physical AI | [API](https://api.aki1k.com/v1/talent/jim-fan) |
| **#19** | [Dan Hendrycks](https://aki1k.com/talent/dan-hendrycks) | `Scientist` | Center for AI Safety (CAIS) | **94.2** | 96.2 | 92.0 | +310% | MMLU Benchmark (Massive Multitask Language Understanding), GELU Activation Function (used across all frontier LLMs), MATH Competition Benchmark, AI Safety Governance Standards | [API](https://api.aki1k.com/v1/talent/dan-hendrycks) |
| **#20** | [Wojciech Zaremba](https://aki1k.com/talent/wojciech-zaremba) | `Research Leader` | OpenAI | **94.0** | 96.0 | 97.0 | +260% | OpenAI Co-Founder, Robotic Hand Dactyl (Rubik's Cube), Neural Programmer-Interpreters, Recurrent Neural Network Regularization | [API](https://api.aki1k.com/v1/talent/wojciech-zaremba) |
| **#21** | [Kaiming He](https://aki1k.com/talent/kaiming-he) | `Scientist` | MIT EECS & CSAIL | **93.8** | 99.7 | 94.0 | +290% | ResNet (Residual Networks - 250k+ Citations), Mask R-CNN, Masked Autoencoders (MAE), PReLU & He Initialization | [API](https://api.aki1k.com/v1/talent/kaiming-he) |
| **#22** | [Pieter Abbeel](https://aki1k.com/talent/pieter-abbeel) | `Physical AI Scientist` | Physical Intelligence & UC Berkeley | **93.6** | 97.5 | 96.0 | +450% | Robot Learning & Apprenticeship Learning, Physical Intelligence (pi-zero), TRPO / PPO Foundations, Berkeley Robot Learning Lab | [API](https://api.aki1k.com/v1/talent/pieter-abbeel) |
| **#23** | [Jeff Dean](https://aki1k.com/talent/jeff-dean) | `Research Leader` | Google DeepMind (Chief Scientist) | **93.5** | 96.0 | 99.5 | +220% | TensorFlow & TPU Distributed Systems, MapReduce & BigTable, Large Scale Deep Learning (DistBelief), Google Brain Co-Founder | [API](https://api.aki1k.com/v1/talent/jeff-dean) |
| **#24** | [Jason Wei](https://aki1k.com/talent/jason-wei) | `Scientist` | OpenAI | **93.3** | 96.5 | 92.5 | +520% | Chain-of-Thought Prompting (CoT), Emergent Abilities in Large Language Models, Instruction Fine-Tuning (FLAN), o1 Reasoning Paradigms | [API](https://api.aki1k.com/v1/talent/jason-wei) |
| **#25** | [Woosuk Kwon](https://aki1k.com/talent/woosuk-kwon) | `Engineer` | vLLM & UC Berkeley | **93.1** | 94.0 | 99.4 | +560% | vLLM Inference Engine, PagedAttention Virtual Memory Management, High-Throughput LLM Serving, Speculative Decoding Systems | [API](https://api.aki1k.com/v1/talent/woosuk-kwon) |
| **#26** | [Chris Olah](https://aki1k.com/talent/chris-olah) | `Scientist` | Anthropic | **92.9** | 98.0 | 90.5 | +380% | Mechanistic Interpretability, Circuits in Neural Networks, Monosemantic Features in Claude, Distill.pub Co-Founder | [API](https://api.aki1k.com/v1/talent/chris-olah) |
| **#27** | [Liang Wenfeng](https://aki1k.com/talent/liang-wenfeng) | `Research Leader` | DeepSeek & High-Flyer | **92.7** | 95.5 | 98.0 | +590% | DeepSeek-V3 Architecture, DeepSeek-R1 Reasoning Open Weights, Multi-Head Latent Attention (MLA), High-Flyer GPU Cluster Engineering | [API](https://api.aki1k.com/v1/talent/liang-wenfeng) |
| **#28** | [Philippe Tillet](https://aki1k.com/talent/philippe-tillet) | `Engineer` | OpenAI | **92.5** | 93.0 | 99.6 | +480% | Triton GPU Programming Language, Custom CUDA Kernel Abstractions, Deep Learning Compiler Optimizations, Flash-Decoding Triton Ports | [API](https://api.aki1k.com/v1/talent/philippe-tillet) |
| **#29** | [Albert Gu](https://aki1k.com/talent/albert-gu) | `Scientist` | CMU & Cartesia | **92.4** | 97.0 | 94.5 | +510% | Mamba State Space Model (SSM), Structured State Spaces (S4), Sub-quadratic Linear Attention, Cartesia Realtime Voice Engine | [API](https://api.aki1k.com/v1/talent/albert-gu) |
| **#30** | [Jared Kaplan](https://aki1k.com/talent/jared-kaplan) | `Scientist` | Anthropic | **92.2** | 97.2 | 92.0 | +310% | Scaling Laws for Neural Language Models, Anthropic Co-Founder & Chief Scientist, Chinchilla Compute-Optimal Pre-training, Theoretical Physics of AI | [API](https://api.aki1k.com/v1/talent/jared-kaplan) |
| **#31** | [John Schulman](https://aki1k.com/talent/john-schulman) | `Research Leader` | Anthropic / Ex-OpenAI | **92.0** | 97.5 | 94.0 | +330% | Proximal Policy Optimization (PPO), Trust Region Policy Optimization (TRPO), ChatGPT Reinforcement Learning Lead, Anthropic Alignment Architect | [API](https://api.aki1k.com/v1/talent/john-schulman) |
| **#32** | [Alec Radford](https://aki1k.com/talent/alec-radford) | `Scientist` | OpenAI | **91.8** | 98.5 | 92.0 | +270% | GPT-1, GPT-2 & GPT-3 Core Architect, CLIP (Contrastive Language-Image Pretraining), Whisper Automatic Speech Recognition, DCGAN (Deep Convolutional GANs) | [API](https://api.aki1k.com/v1/talent/alec-radford) |
| **#33** | [Jakub Pachocki](https://aki1k.com/talent/jakub-pachocki) | `Research Leader` | OpenAI (Chief Scientist) | **91.6** | 96.8 | 97.5 | +360% | GPT-4 Core Architecture Lead, OpenAI o1 Reasoning System, OpenAI Five Dota 2 Reinforcement Learning, Frontier Scaling Algorithms | [API](https://api.aki1k.com/v1/talent/jakub-pachocki) |
| **#34** | [Mark Chen](https://aki1k.com/talent/mark-chen) | `Research Leader` | OpenAI (SVP Research) | **91.4** | 96.2 | 97.0 | +340% | Codex / GitHub Copilot Foundation, DALL-E 2 & Image Generation, Sora Video Model Research Lead, Multimodal Model Systems | [API](https://api.aki1k.com/v1/talent/mark-chen) |
| **#35** | [Daya Guo](https://aki1k.com/talent/daya-guo) | `Engineer` | DeepSeek | **91.2** | 94.8 | 98.2 | +570% | DeepSeek-Coder Architecture, DeepSeekMath Reinforcement Learning, DeepSeek-V3 Core Training Engineer, Multi-Token Prediction Optimization | [API](https://api.aki1k.com/v1/talent/daya-guo) |
| **#36** | [Jan Leike](https://aki1k.com/talent/jan-leike) | `Scientist` | Anthropic / Ex-OpenAI | **91.0** | 96.5 | 91.5 | +310% | Superalignment Research Lead, InstructGPT RLHF Co-Author, AI Safety by Debate, Anthropic Alignment Science | [API](https://api.aki1k.com/v1/talent/jan-leike) |
| **#37** | [Oriol Vinyals](https://aki1k.com/talent/oriol-vinyals) | `Research Leader` | Google DeepMind (VP Research) | **90.8** | 97.0 | 94.5 | +280% | Seq2Seq with Attention, AlphaStar StarCraft II Grandmaster AI, Gemini Core Multimodal Lead, Matching Networks Few-Shot Learning | [API](https://api.aki1k.com/v1/talent/oriol-vinyals) |
| **#38** | [David Silver](https://aki1k.com/talent/david-silver) | `Scientist` | Google DeepMind | **90.6** | 98.0 | 91.0 | +230% | AlphaGo Mastermind, AlphaZero General Game Playing, MuZero Model-Based RL, Deep Reinforcement Learning Foundations | [API](https://api.aki1k.com/v1/talent/david-silver) |
| **#39** | [Soumith Chintala](https://aki1k.com/talent/soumith-chintala) | `Engineer` | Meta FAIR | **90.4** | 92.5 | 99.0 | +290% | PyTorch Framework Creator, TorchVision Library, DCGAN Co-Author, High-Performance Open Compute Systems | [API](https://api.aki1k.com/v1/talent/soumith-chintala) |
| **#40** | [Percy Liang](https://aki1k.com/talent/percy-liang) | `Scientist` | Stanford HAI & DSPy | **90.2** | 97.5 | 91.5 | +410% | DSPy Declarative Self-Improving Programs, HELM Holistic Evaluation of Language Models, Foundation Model Benchmark Standards, Semantic Parsing Frameworks | [API](https://api.aki1k.com/v1/talent/percy-liang) |
| **#41** | [Christopher Manning](https://aki1k.com/talent/christopher-manning) | `Scientist` | Stanford University NLP | **90.0** | 98.0 | 89.0 | +240% | GloVe Word Embeddings, Stanford CoreNLP Suite, Tree-Structured Recursive Neural Networks, Foundations of Statistical NLP | [API](https://api.aki1k.com/v1/talent/christopher-manning) |
| **#42** | [Danqi Chen](https://aki1k.com/talent/danqi-chen) | `Scientist` | Princeton University NLP | **89.8** | 96.5 | 91.0 | +430% | Dense Passage Retrieval (DPR) for RAG, DrQA Neural Question Answering, Fast Exact Search in Vector DBs, Princeton NLP Lab Director | [API](https://api.aki1k.com/v1/talent/danqi-chen) |
| **#43** | [Song Han](https://aki1k.com/talent/song-han) | `Engineer` | MIT EECS & OmniML | **89.6** | 93.5 | 98.5 | +460% | Deep Compression (Pruning & Quantization), TinyML & TinyLlama Frameworks, SmoothQuant 8-bit Weight/Activation, Efficient Spatially-Sparse Computing | [API](https://api.aki1k.com/v1/talent/song-han) |
| **#44** | [Chelsea Finn](https://aki1k.com/talent/chelsea-finn) | `Physical AI Scientist` | Stanford University | **89.4** | 96.0 | 93.0 | +390% | Model-Agnostic Meta-Learning (MAML), BridgeData Robotics Manipulation, Generalist Vision-Language-Action Models, Dexterous Robotic Manipulation | [API](https://api.aki1k.com/v1/talent/chelsea-finn) |
| **#45** | [Sergey Levine](https://aki1k.com/talent/sergey-levine) | `Physical AI Scientist` | UC Berkeley Robotic Learning Lab | **89.2** | 97.0 | 92.5 | +370% | Guided Policy Search (GPS), QT-Opt Scalable Robotic Manipulation, Offline Reinforcement Learning (CQL), Open X-Embodiment Dataset | [API](https://api.aki1k.com/v1/talent/sergey-levine) |
| **#46** | [Bo Peng](https://aki1k.com/talent/bo-peng) | `Engineer` | RWKV Foundation | **89.0** | 94.0 | 97.5 | +480% | RWKV RNN-Transformer Architecture, O(1) Linear Time Inference, Sub-quadratic Sequence Modeling, Open Multi-Lingual Weights | [API](https://api.aki1k.com/v1/talent/bo-peng) |
| **#47** | [Shuming Ma](https://aki1k.com/talent/shuming-ma) | `Scientist` | Microsoft Research Asia | **88.8** | 95.5 | 94.0 | +500% | BitNet 1-bit LLM Architecture, The Era of 1-bit LLMs, BitLinear Matrix Multiplication, Zero-Multiplication Neural Compute | [API](https://api.aki1k.com/v1/talent/shuming-ma) |
| **#48** | [Matei Zaharia](https://aki1k.com/talent/matei-zaharia) | `Engineer` | Databricks & UC Berkeley | **88.6** | 92.0 | 98.8 | +350% | Apache Spark Distributed Framework, DSPy Co-Author & Optimizer, MLflow Machine Learning Lifecycle, Databricks DBRX Architecture | [API](https://api.aki1k.com/v1/talent/matei-zaharia) |
| **#49** | [Ion Stoica](https://aki1k.com/talent/ion-stoica) | `Engineer` | Anyscale & UC Berkeley | **88.4** | 91.5 | 98.5 | +330% | Ray Distributed AI Execution Framework, vLLM Co-Author, Apache Spark Co-Founder, Anyscale Endpoints | [API](https://api.aki1k.com/v1/talent/ion-stoica) |
| **#50** | [Dawn Song](https://aki1k.com/talent/dawn-song) | `Scientist` | UC Berkeley | **88.2** | 96.0 | 91.0 | +310% | AI Security & Adversarial Robustness, Privacy-Preserving Machine Learning, Oasis Labs Blockchain AI, Data Provenance Standards | [API](https://api.aki1k.com/v1/talent/dawn-song) |
| **#51** | [Thomas Wolf](https://aki1k.com/talent/thomas-wolf) | `Engineer` | Hugging Face (Co-Founder & CSO) | **88.0** | 93.0 | 97.0 | +320% | Transformers Python Library, BigScience BLOOM Open LLM, Hugging Face Datasets, Open Science Collaboration Leader | [API](https://api.aki1k.com/v1/talent/thomas-wolf) |
| **#52** | [Harrison Chase](https://aki1k.com/talent/harrison-chase) | `Engineer` | LangChain (CEO & Co-Founder) | **87.8** | 90.0 | 98.0 | +450% | LangChain Framework, LangGraph Multi-Agent Orchestration, LangSmith Observability Suite, Enterprise Agentic Pipelines | [API](https://api.aki1k.com/v1/talent/harrison-chase) |
| **#53** | [Jerry Liu](https://aki1k.com/talent/jerry-liu) | `Engineer` | LlamaIndex (CEO & Co-Founder) | **87.6** | 90.5 | 97.5 | +440% | LlamaIndex Context Augmentation Framework, Advanced RAG Retrieval Algorithms, LlamaParse PDF/Document Engine, Graph RAG Implementations | [API](https://api.aki1k.com/v1/talent/jerry-liu) |
| **#54** | [Jeremy Howard](https://aki1k.com/talent/jeremy-howard) | `Engineer` | fast.ai & Answer.AI | **87.4** | 91.0 | 97.0 | +310% | ULMFiT (Universal Language Model Fine-tuning), fastai Deep Learning Library, Democratized Global AI Education, Answer.AI Open R&D Lab | [API](https://api.aki1k.com/v1/talent/jeremy-howard) |
| **#55** | [Paul Christiano](https://aki1k.com/talent/paul-christiano) | `Scientist` | Alignment Research Center (ARC) | **87.2** | 97.0 | 88.0 | +340% | RLHF Foundation Paper (Christiano et al. 2017), Eliciting Latent Knowledge (ELK), ARC Model Evaluation Standards, Iterated Amplification | [API](https://api.aki1k.com/v1/talent/paul-christiano) |
| **#56** | [Stuart Russell](https://aki1k.com/talent/stuart-russell) | `Scientist` | UC Berkeley | **87.0** | 98.0 | 86.0 | +210% | AI: A Modern Approach (Standard Textbook), Inverse Reinforcement Learning (IRL), Human Compatible AI Governance, Center for Human-Compatible AI | [API](https://api.aki1k.com/v1/talent/stuart-russell) |
| **#57** | [Max Tegmark](https://aki1k.com/talent/max-tegmark) | `Scientist` | MIT & Future of Life Institute | **86.8** | 96.5 | 87.5 | +260% | AI Safety Open Letters & Standards, Mechanistic Interpretability of Representation, Life 3.0 Best-Selling Framework, Future of Life Institute Founder | [API](https://api.aki1k.com/v1/talent/max-tegmark) |
| **#58** | [Noam Brown](https://aki1k.com/talent/noam-brown) | `Scientist` | OpenAI | **86.6** | 96.8 | 92.5 | +490% | Libratus & Pluribus (No-Limit Poker Superhuman AI), Cicero (Diplomacy Strategic Dialogue AI), OpenAI o1 Multi-Step Reasoning Lead, Search-at-Inference Algorithms | [API](https://api.aki1k.com/v1/talent/noam-brown) |
| **#59** | [Sheng Shen](https://aki1k.com/talent/sheng-shen) | `Engineer` | Together AI & UC Berkeley | **86.4** | 93.5 | 97.0 | +480% | FlashDecoding Long-Context Inference, Hessian-Free Neural Optimization, LLM Model Merging Algorithms, Together Inference Engine Kernel Lead | [API](https://api.aki1k.com/v1/talent/sheng-shen) |
| **#60** | [Zhuohan Li](https://aki1k.com/talent/zhuohan-li) | `Engineer` | vLLM & UC Berkeley | **86.2** | 92.5 | 97.5 | +520% | PagedAttention Co-Author, Alpa Distributed Auto-Parallelization, vLLM Prefix Caching & Chunked Prefill, GPU Memory Scheduling | [API](https://api.aki1k.com/v1/talent/zhuohan-li) |
| **#61** | [Armand Joulin](https://aki1k.com/talent/armand-joulin) | `Scientist` | Apple AI Research / Ex-Meta | **86.0** | 95.5 | 92.0 | +280% | FastText Word Representation Library, LLaMA 1 Architecture Core Author, Self-Supervised Visual Transformers (DINO), Mobile AI Research | [API](https://api.aki1k.com/v1/talent/armand-joulin) |
| **#62** | [Timothée Lacroix](https://aki1k.com/talent/timothee-lacroix) | `Engineer` | Mistral AI (CTO & Co-Founder) | **85.8** | 93.0 | 97.0 | +380% | Mistral AI Infrastructure & Training Stack, LLaMA 1 Distributed Training Lead at Meta, Tensor Parallelism Optimizations, High-Throughput French Compute | [API](https://api.aki1k.com/v1/talent/timothee-lacroix) |
| **#63** | [Barret Zoph](https://aki1k.com/talent/barret-zoph) | `Scientist` | Google DeepMind / Ex-OpenAI | **85.6** | 95.8 | 92.0 | +310% | Neural Architecture Search (NAS), ST-MoE Stable MoE Routing, Mixture-of-Experts Scaling Laws, OpenAI Post-Training Research | [API](https://api.aki1k.com/v1/talent/barret-zoph) |
| **#64** | [Colin Raffel](https://aki1k.com/talent/colin-raffel) | `Scientist` | University of Toronto & Cohere | **85.4** | 96.0 | 90.0 | +290% | T5 (Text-to-Text Transfer Transformer), C4 Clean Colossal Common Crawl Dataset, BigScience Project Leadership, Open Language Modeling Evaluations | [API](https://api.aki1k.com/v1/talent/colin-raffel) |
| **#65** | [Trieu Trinh](https://aki1k.com/talent/trieu-trinh) | `Scientist` | Google DeepMind | **85.2** | 96.2 | 89.5 | +470% | AlphaGeometry Olympiad-Level Geometry AI, AlphaProof Formal Mathematical Reasoning, Synthetic Data Generation for Math, Neuro-Symbolic Reasoning | [API](https://api.aki1k.com/v1/talent/trieu-trinh) |
| **#66** | [Pushmeet Kohli](https://aki1k.com/talent/pushmeet-kohli) | `Research Leader` | Google DeepMind (VP AI Science) | **85.0** | 96.0 | 91.5 | +320% | AlphaFold 2 & 3 Program Leadership, FunSearch Mathematical Discovery, GraphCast AI Weather Prediction, GNoME Crystal Structure Discovery | [API](https://api.aki1k.com/v1/talent/pushmeet-kohli) |
| **#67** | [Alex Graves](https://aki1k.com/talent/alex-graves) | `Scientist` | Independent / Ex-DeepMind | **84.8** | 97.2 | 88.0 | +210% | Connectionist Temporal Classification (CTC), Neural Turing Machines (NTM), Differentiable Neural Computer (DNC), Recurrent Neural Network Transducer | [API](https://api.aki1k.com/v1/talent/alex-graves) |
| **#68** | [Koray Kavukcuoglu](https://aki1k.com/talent/koray-kavukcuoglu) | `Research Leader` | Google DeepMind (CTO) | **84.6** | 95.5 | 93.0 | +270% | DeepMind Core Research Roadmap, DQN Reinforcement Learning Co-Author, AlphaGo Infrastructure Oversight, Gemini Architecture Governance | [API](https://api.aki1k.com/v1/talent/koray-kavukcuoglu) |
| **#69** | [Zoubin Ghahramani](https://aki1k.com/talent/zoubin-ghahramani) | `Scientist` | University of Cambridge & Google | **84.4** | 96.5 | 88.0 | +200% | Probabilistic Machine Learning Foundations, Gaussian Processes for ML, Bayesian Nonparametrics, Uber AI Labs Founder | [API](https://api.aki1k.com/v1/talent/zoubin-ghahramani) |
| **#70** | [Quoc Le](https://aki1k.com/talent/quoc-le) | `Scientist` | Google DeepMind | **84.2** | 96.8 | 90.0 | +250% | Sequence-to-Sequence Co-Inventor, AutoML & Neural Architecture Search, Doc2Vec & Paragraph Vectors, Google Brain Founding Member | [API](https://api.aki1k.com/v1/talent/quoc-le) |
| **#71** | [Daniela Amodei](https://aki1k.com/talent/daniela-amodei) | `Research Leader` | Anthropic (President & Co-Founder) | **84.0** | 92.0 | 94.0 | +340% | Anthropic Institutional Scale, Safety-First Model Governance, Frontier AI Safety Standards, OpenAI VP Operations | [API](https://api.aki1k.com/v1/talent/daniela-amodei) |
| **#72** | [Jack Clark](https://aki1k.com/talent/jack-clark) | `Research Leader` | Anthropic (Co-Founder & Policy) | **83.8** | 91.5 | 93.5 | +330% | Import AI Weekly Newsletter, Compute Index Tracking, OECD AI Governance Group, National AI Research Resource (NAIRR) | [API](https://api.aki1k.com/v1/talent/jack-clark) |
| **#73** | [Sam McCandlish](https://aki1k.com/talent/sam-mccandlish) | `Scientist` | Anthropic | **83.6** | 95.0 | 91.0 | +320% | An Empirical Model of Large-Batch Training, Scaling Laws for Autoregressive Generative Modeling, Anthropic Interpretability Foundations, Critical Batch Size Analysis | [API](https://api.aki1k.com/v1/talent/sam-mccandlish) |
| **#74** | [Tom Brown](https://aki1k.com/talent/tom-brown) | `Engineer` | Anthropic / Ex-OpenAI | **83.4** | 93.0 | 95.0 | +300% | GPT-3 Lead Author (Language Models are Few-Shot Learners), Large Scale Distributed Training at OpenAI, Anthropic Claude Infrastructure, Few-Shot Evaluation Paradigms | [API](https://api.aki1k.com/v1/talent/tom-brown) |
| **#75** | [Amanda Askell](https://aki1k.com/talent/amanda-askell) | `Scientist` | Anthropic | **83.2** | 94.5 | 88.0 | +310% | Constitutional AI Co-Author, Claude Character & Behavioral Training, Philosophy of AI Minds, RLHF Evaluation Methodologies | [API](https://api.aki1k.com/v1/talent/amanda-askell) |
| **#76** | [Mira Murati](https://aki1k.com/talent/mira-murati) | `Research Leader` | Thinking Machines Lab / Ex-OpenAI | **83.0** | 92.5 | 95.5 | +380% | ChatGPT, GPT-4 & GPT-4o Release Lead, DALL-E & Sora Product Direction, Leap Motion VP Product, Thinking Machines Frontier Lab | [API](https://api.aki1k.com/v1/talent/mira-murati) |
| **#77** | [Greg Brockman](https://aki1k.com/talent/greg-brockman) | `Engineer` | OpenAI (President & Co-Founder) | **82.8** | 90.0 | 97.5 | +260% | OpenAI Supercomputing Infrastructure, Stripe CTO & Core Payments Engine, OpenAI Gym & Universe Foundations, Azure-OpenAI Multi-Cluster Systems | [API](https://api.aki1k.com/v1/talent/greg-brockman) |
| **#78** | [Liam Fedus](https://aki1k.com/talent/liam-fedus) | `Scientist` | OpenAI (VP Post-Training) | **82.6** | 95.2 | 92.0 | +390% | Switch Transformers (1.6 Trillion Parameter MoE), OpenAI Post-Training Architecture Lead, Reasoning Model RL Pipeline, ManiSkill Robotics Benchmarks | [API](https://api.aki1k.com/v1/talent/liam-fedus) |
| **#79** | [Shengjia Zhao](https://aki1k.com/talent/shengjia-zhao) | `Scientist` | OpenAI (VP Research) | **82.4** | 95.0 | 91.5 | +370% | Information-Theoretic AI Foundations, Variational Loss Bounds, OpenAI Reasoning & Multi-Modal Models, Stanford AI PhD with Stefano Ermon | [API](https://api.aki1k.com/v1/talent/shengjia-zhao) |
| **#80** | [Nick Turley](https://aki1k.com/talent/nick-turley) | `Research Leader` | OpenAI | **82.2** | 91.0 | 94.0 | +350% | ChatGPT Product Strategy & Voice Mode, Advanced Data Analysis Engine, Custom GPTs Ecosystem, User-Facing LLM Interaction Models | [API](https://api.aki1k.com/v1/talent/nick-turley) |
| **#81** | [Bob McGrew](https://aki1k.com/talent/bob-mcgrew) | `Research Leader` | Independent / Ex-OpenAI | **82.0** | 92.0 | 94.5 | +240% | OpenAI VP Research (Execution), Palantir Director of Engineering, OpenAI Five Infrastructure Lead, Frontier Alignment Deployment | [API](https://api.aki1k.com/v1/talent/bob-mcgrew) |
| **#82** | [Jerry Tworek](https://aki1k.com/talent/jerry-tworek) | `Research Leader` | OpenAI (VP Research) | **81.8** | 94.0 | 93.0 | +320% | InstructGPT & Human Feedback RL Lead, Codex Evaluation Framework, Mathematical Reasoning Evaluation, OpenAI Alignment Science | [API](https://api.aki1k.com/v1/talent/jerry-tworek) |
| **#83** | [Joelle Pineau](https://aki1k.com/talent/joelle-pineau) | `Research Leader` | Meta FAIR & McGill University | **81.6** | 95.0 | 89.0 | +270% | Reproducibility in Machine Learning, Partially Observable Markov Decision Processes (POMDPs), Meta FAIR VP AI Research, Mila Founding Member | [API](https://api.aki1k.com/v1/talent/joelle-pineau) |
| **#84** | [Ross Girshick](https://aki1k.com/talent/ross-girshick) | `Scientist` | Meta FAIR | **81.4** | 96.5 | 88.0 | +210% | R-CNN, Fast R-CNN & Faster R-CNN, Detectron Computer Vision Suite, Deformable Part Models (DPM), Feature Pyramid Networks (FPN) | [API](https://api.aki1k.com/v1/talent/ross-girshick) |
| **#85** | [Edward Yang](https://aki1k.com/talent/edward-yang) | `Engineer` | Meta PyTorch Core | **81.2** | 89.0 | 98.5 | +340% | PyTorch Core Dispatcher & TorchDynamo, TorchScript JIT Compiler, Functional PyTorch Tensor Subsystems, ezyang GHC & PyTorch Architecture | [API](https://api.aki1k.com/v1/talent/edward-yang) |
| **#86** | [Michael Auli](https://aki1k.com/talent/michael-auli) | `Scientist` | Meta FAIR | **81.0** | 95.0 | 90.0 | +280% | wav2vec & wav2vec 2.0 Speech Models, fairseq Sequence Modeling Toolkit, Adaptive Computation in Transformers, High-Performance Open Speech | [API](https://api.aki1k.com/v1/talent/michael-auli) |
| **#87** | [Naman Goyal](https://aki1k.com/talent/naman-goyal) | `Engineer` | Meta AI | **80.8** | 91.5 | 96.0 | +360% | LLaMA 1, 2 & 3 Core Engineering Lead, Megatron-LM at Meta Scale, RoPE (Rotary Position Embedding) Scaling, Distributed Sharded Checkpointing | [API](https://api.aki1k.com/v1/talent/naman-goyal) |
| **#88** | [Bryan Catanzaro](https://aki1k.com/talent/bryan-catanzaro) | `Engineer` | NVIDIA (VP Applied Deep Learning) | **80.6** | 91.0 | 97.0 | +330% | Megatron-LM Large Scale Distributed Framework, cuDNN Deep Neural Network Library, Deep Speech GPU Architecture, TensorRT Inference Acceleration | [API](https://api.aki1k.com/v1/talent/bryan-catanzaro) |
| **#89** | [Stan Birchfield](https://aki1k.com/talent/stan-birchfield) | `Physical AI Scientist` | NVIDIA Research | **80.4** | 94.0 | 91.0 | +300% | NVIDIA Isaac Sim Synthetic Data, Kanade-Lucas-Tomasi (KLT) Tracker, Synthetic Perception for Humanoid Robotics, Spatial AI Benchmarks | [API](https://api.aki1k.com/v1/talent/stan-birchfield) |
| **#90** | [Jonah Alben](https://aki1k.com/talent/jonah-alben) | `Engineer` | NVIDIA (SVP GPU Engineering) | **80.2** | 88.0 | 98.8 | +350% | Hopper H100 & Blackwell B200 GPU Architectures, Transformer Engine Hardware Logic, NVLink 5 High-Bandwidth Interconnect, CUDA Core Microarchitecture | [API](https://api.aki1k.com/v1/talent/jonah-alben) |
| **#91** | [Bill Dally](https://aki1k.com/talent/bill-dally) | `Research Leader` | NVIDIA (Chief Scientist & SVP) | **80.0** | 95.0 | 95.0 | +220% | Interconnection Networks & Routers, Stream Processing Architectures, NVIDIA Research Lab Leadership, Hardware Accelerated AI Foundations | [API](https://api.aki1k.com/v1/talent/bill-dally) |
| **#92** | [Yuke Zhu](https://aki1k.com/talent/yuke-zhu) | `Physical AI Scientist` | UT Austin & NVIDIA GEAR Lab | **79.8** | 95.0 | 91.0 | +410% | Robomimic Benchmark & Imitation Learning, VIOLA Visual Robot Manipulation, Project GR0OT Co-Lead, Simulation-to-Real Transfer | [API](https://api.aki1k.com/v1/talent/yuke-zhu) |
| **#93** | [Ruslan Salakhutdinov](https://aki1k.com/talent/ruslan-salakhutdinov) | `Scientist` | CMU (Former Apple Director AI) | **79.6** | 96.0 | 87.0 | +210% | Deep Boltzmann Machines, Multimodal Representation Learning, Apple Foundation AI Research, Deep Generative Models | [API](https://api.aki1k.com/v1/talent/ruslan-salakhutdinov) |
| **#94** | [Jitendra Malik](https://aki1k.com/talent/jitendra-malik) | `Scientist` | UC Berkeley & Meta FAIR | **79.4** | 97.0 | 86.0 | +200% | Normalized Cuts Computer Vision, Bilateral Filtering Foundations, Ecological Approach to Perception, Generations of AI Leaders Mentored | [API](https://api.aki1k.com/v1/talent/jitendra-malik) |
| **#95** | [Trevor Darrell](https://aki1k.com/talent/trevor-darrell) | `Scientist` | UC Berkeley (BAIR Director) | **79.2** | 96.0 | 87.5 | +240% | Berkeley AI Research (BAIR) Director, Caffe Deep Learning Framework Co-Creator, Zero-Shot Object Recognition, Autonomous Driving Perception | [API](https://api.aki1k.com/v1/talent/trevor-darrell) |
| **#96** | [Yang You](https://aki1k.com/talent/yang-you) | `Engineer` | National University of Singapore & Colossal-AI | **79.0** | 92.5 | 96.0 | +460% | Colossal-AI Distributed Systems, LAMB Optimizer (Large Batch Training for BERT), LARS Optimizer (ImageNet in Minutes), NUS AI Chair | [API](https://api.aki1k.com/v1/talent/yang-you) |
| **#97** | [Xiao Bi](https://aki1k.com/talent/xiao-bi) | `Engineer` | DeepSeek | **78.8** | 93.0 | 96.0 | +550% | DeepSeek-R1 Cold Start RL Recipes, Multi-Head Latent Attention Kernel Implementations, DeepSeek MoE Load Balancing Strategies, Custom DualPipe Parallelism | [API](https://api.aki1k.com/v1/talent/xiao-bi) |
| **#98** | [Charlie Snell](https://aki1k.com/talent/charlie-snell) | `Rising Star` | UC Berkeley (BAIR) | **78.6** | 93.5 | 92.0 | +530% | Inference-Time Compute Scaling Laws, Diffusion Language Modeling Analysis, Search vs Pretraining Compute Trade-offs, Agentic Loop Optimizations | [API](https://api.aki1k.com/v1/talent/charlie-snell) |
| **#99** | [Sewon Min](https://aki1k.com/talent/sewon-min) | `Rising Star` | University of Washington | **78.4** | 94.0 | 89.0 | +490% | Rethinking the Role of Demonstrations in In-Context Learning, Nonparametric Language Models, Dense Retrieval Evaluation, Direct Preference Optimization Studies | [API](https://api.aki1k.com/v1/talent/sewon-min) |
| **#100** | [Eric Wallace](https://aki1k.com/talent/eric-wallace) | `Rising Star` | UC Berkeley | **78.2** | 94.5 | 88.0 | +480% | Universal Adversarial Triggers for NLP, Data Extraction from Frontier LLMs, Poisoning Web-Scale Training Data, Safety Red-Teaming Automated Suites | [API](https://api.aki1k.com/v1/talent/eric-wallace) |

---

## Live AI Talent Transfer Window & Lab Migration Feed

| Transfer ID | Individual | Source Lab | Destination Lab | Role Title | Effective Date | Strategic Impact Summary |
| :--- | :--- | :--- | :--- | :--- | :---: | :--- |
| `TR-2026-ILYA-SSI` | [Ilya Sutskever](https://aki1k.com/talent/ilya-sutskever) | **OpenAI** | **Safe Superintelligence (SSI)** | Co-Founder & Chief Scientist | `2024-06-19` | Founded dedicated SSI lab targeting straight-shot Safe Superintelligence without commercial product distractions. |
| `TR-2026-NOAM-DEEPMIND` | [Noam Shazeer](https://aki1k.com/talent/noam-shazeer) | **Character.AI** | **Google DeepMind** | Technical Lead Gemini | `2024-08-16` | Returned to Google in $2.7B licensing & talent acquisition deal to spearhead Gemini next-gen Transformer models. |
| `TR-2026-KARPATHY-EUREKA` | [Andrej Karpathy](https://aki1k.com/talent/andrej-karpathy) | **OpenAI** | **Eureka Labs** | Founder & CEO | `2024-07-16` | Launched Eureka Labs to build AI-native educational systems combining LLMs with human curriculum. |
| `TR-2026-SCHULMAN-ANTHROPIC` | [John Schulman](https://aki1k.com/talent/john-schulman) | **OpenAI** | **Anthropic** | Alignment Research Lead | `2024-08-05` | OpenAI Co-Founder and PPO architect joined Anthropic to focus hands-on on empirical alignment and reinforcement learning. |
| `TR-2026-LEIKE-ANTHROPIC` | [Jan Leike](https://aki1k.com/talent/jan-leike) | **OpenAI** | **Anthropic** | Chief Superalignment Scientist | `2024-05-28` | Former OpenAI Superalignment co-lead migrated to Anthropic to build institutional safety benchmarks and scalability tests. |
| `TR-2026-MURATI-THINKING` | [Mira Murati](https://aki1k.com/talent/mira-murati) | **OpenAI** | **Thinking Machines Lab** | Founder & CEO | `2024-09-25` | Former OpenAI CTO announced new independent AI laboratory focusing on proprietary multimodal foundation models. |
| `TR-2026-ABBEEL-PI` | [Pieter Abbeel](https://aki1k.com/talent/pieter-abbeel) | **UC Berkeley** | **Physical Intelligence (π)** | Co-Founder & Scientific Advisor | `2024-11-04` | Launched Physical Intelligence with $400M seed from Bezos & Thrive Capital to build universal robotic foundation models (pi-zero). |
| `TR-2026-TRI-DAO-TOGETHER` | [Tri Dao](https://aki1k.com/talent/tri-dao) | **Stanford University** | **Together AI & Princeton** | Chief Scientist & Assistant Professor | `2024-01-15` | FlashAttention creator joined Together AI as Chief Scientist while establishing Princeton systems lab for hardware-aware architectures. |
| `TR-2026-LLION-SAKANA` | [Llion Jones](https://aki1k.com/talent/llion-jones) | **Google** | **Sakana AI** | Co-Founder & CTO | `2023-08-17` | Pioneered transformer architecture author launched Sakana AI in Tokyo focusing on evolutionary model merging and nature-inspired foundation intelligence. |
| `TR-2026-AIDAN-COHERE` | [Aidan Gomez](https://aki1k.com/talent/aidan-gomez) | **Google Brain** | **Cohere** | Co-Founder & CEO | `2023-06-01` | Co-author of 'Attention Is All You Need' founded enterprise foundation lab Cohere, deploying state-of-the-art Command-R+ models. |
| `TR-2026-JAKOB-INCEPTIVE` | [Jakob Uszkoreit](https://aki1k.com/talent/jakob-uszkoreit) | **Google** | **Inceptive** | Co-Founder & CEO | `2023-01-10` | Conceived initial multi-head attention Transformer proposal; founded Inceptive deploying deep learning for programmable RNA therapeutics. |
| `TR-2026-ARMAND-EXTRAPOLENCE` | [Armand Joulin](https://aki1k.com/talent/armand-joulin) | **Meta FAIR** | **Extrapolence & DeepMind** | Co-Founder & VP Research | `2024-03-12` | Llama 1/2 co-lead and FastText creator transitioned from Meta to launch frontier research lab Extrapolence focusing on long-horizon reasoning systems. |

---

## Machine Access & MCP Endpoints

- **REST API - Talent 100:** `GET https://api.aki1k.com/v1/talent?limit=100`
- **REST API - Person Detail:** `GET https://api.aki1k.com/v1/talent/:slug`
- **REST API - Transfers Feed:** `GET https://api.aki1k.com/v1/talent/transfers`
- **Model Context Protocol (MCP):** `POST https://api.aki1k.com/v1/mcp` (Tool: `query_ai_talent`)
- **JSON Manifest:** `https://aki1k.com/talent.json`
- **Contact & Corrections:** `team@aki1k.com`