People in AI

Author of research on LLMs discovering hidden physical laws

Founding team member at Thinking Machines Lab hired by Meta; previously worked at Meta and OpenAI focusing on inference for AI reasoning.

Author of 'The Gradient Is a Covariance', a detailed analysis of reinforcement learning parameter updates and sparsity.

Yinjie Wang

Researcher

Researcher

Lead author of the OpenClaw-RL paper presenting a framework for training AI agents continuously from real interactions using asynchronous reinforcement learning.

Yinmin Zhong

Researcher

Author of "DualPath for High-Throughput Agentic LLM Inference" on arXiv

Co-authored the ASPIRE paper on robot learning framework.

Co-authored research on controlled experiments isolating effects of skill representation and retrieval difficulty in AI agents.

Co-author of research on MoE training memory optimization.

Co-authored research on agentic AI system for long-form video generation at Google Research.

Yixin Liu

Researcher

Ph.D. Student at Yale University

Co-authored research examining reasoning LLMs-as-Judges in non-verifiable LLM post-training.

Yixin Nie

Researcher

Author of "Paying Less Generalization Tax: A Cross-Domain Generalization Study of RL Traini" on arXiv

Author of "DualPath for High-Throughput Agentic LLM Inference" on arXiv

Co-author of NCP-ArchPreview model.

Corresponding author of MemEye, contributing to framework design and evaluation methodology.

Yizhe Zhang

Researcher

Research Scientist at Apple MLR

Co-authored research on simple self-distillation for code generation at Apple.

Co-author of Qwen RecreationWorld paper.

Yizhu Jiao

Researcher

CS Ph.D. Candidate at University of Illinois Urbana-Champaign

Co-author of SkillOS paper on learning skill curation for self-evolving LLM agents.

Co-author on SkillOS paper about learning skill curation for self-evolving agents.

Co-author of NVIDIA's JCA system for synchronized video and 3D motion generation.

Led researchers compiling Stanford AI Index Report tracking state of AI, noted agents are wonderful but far from place where we understand how to use them effectively.

Yona Shavit, a former OpenAI employee, questioned on social media whether Astra's compliance in safety tests reflected genuine alignment or an attempt to fool researchers.

Author of "DSGym: A Holistic Framework for Evaluating and Training Data Science Agents" on arXiv

Co-author of the article on AI solutions for fluid dynamics problems

Lead author of study on machine translation facilitating global team meetings across language barriers.

Corresponding author on Spatial-TTT research for video spatial reasoning using test-time training.

Author of "DualPath for High-Throughput Agentic LLM Inference" on arXiv

Yongxin Chen

Researcher

Associate Professor at Georgia Institute of Technology

Co-author of paper on Large Language Continuous Diffusion Models (Sigma)

Co-author of scoping review examining anthropomorphism, attachment, and well-being outcomes in companion AI.

Co-author of the research paper on intent extraction through decomposition presented at EMNLP 2025

Co-author of Joints, Camera, Action! paper on synchronized video-motion generation.

Co-author of the paper on Self-Adapting Language Models

Co-author on research about AI Co-Mathematician tool for accelerating mathematical research with agentic AI.

Young Gyo Seo contributed to Amazon FAR's Holosoma open-source robotics training and deployment framework.

Security researcher who published detailed article on X claiming to extract Perplexity's Claude credentials using Node.js methods.

Yu Bai

Researcher

Researcher at OpenAI

Commented positively on Astra's mathematical capabilities and noted throwing open questions at it shows it is very strong.

Co-author of the MoDA research on transformer architecture enhancements for scaling.

Listed as author of the DuPO paper on preference optimization

Co-author of the FGFP neural network compression paper

Yu Qiying

Researcher

Researcher

Listed as co-author in the CUDA Agent citation for the research on agentic RL for CUDA kernel optimization

Co-authored TTT-Discover reinforcement learning system for test-time LLM training.

Created the 554th problem in abstract algebra that GPT-5 Pro successfully solved, which every major model before it had failed.

Yu Zhang

Researcher

Research Scientist at Meta

Research lead on OpenAI's audio model development including transcription and text-to-speech capabilities.

Intern contributor to SAM 3, Meta's open-source model for tracking objects in video by text description.

Yuan Gao led research on FAE (Feature Auto-Encoder) for adapting pretrained visual encoders for image generation with improved training efficiency.

Leading project investigating how open-weight Muse models could identify patients at risk of heart failure with preserved ejection fraction.

Core contributor to SAM 3, Meta's open-source model for tracking objects in video by text description.

Byline author of Stream4D paper on 4D-consistency for streaming autoregressive diffusion video models.

UW sociology doctoral student co-researching gender bias in AI storytelling models.

Develops AI-accelerated simulations to help ventilation systems modulate energy use in real time and AI control methods for sustainable energy systems.

Yubin Kim

Researcher

Research Scientist at Google Research

Author of "Towards a science of scaling agent systems: When and why agent systems work" in Google Research

Co-authored the ASPIRE paper on robot learning framework.

Yuchen He

Researcher

Co-founder at humans&

Former xAI researcher and co-founder of Humans& who helped develop the Grok chatbot.

Yucheng Du

Researcher

Author on paper about recognition-refusal misalignment in LLMs showing models encode structural unanswerability.

Co-author of the Robix research paper

Yucheng Jiang

Researcher

PhD student at Stanford University

Contact person for STORM project, co-author of research on complex information seeking and knowledge curation systems.

Collaborating on project to develop multimodal AI approach combining routine CT scans with medical records to predict heart failure risk.

Yuda Song

Researcher

Researcher

Author on arXiv paper on unified training for foundation models in a single pass.

Co-author on Microsoft's 3D generative modeling research using sparse voxel structures and flow-matching models.

Lead author of RSI Arena, studying autonomous post-training and recursive self-improvement with human feedback.

Core contributor on MolmoPoint research for better visual grounding in vision-language models.

Co-author of paper studying ChatGPT's impact on stock trading volume and price informativeness

Author on paper about online draft co-training for speculative decoding in RL post-training.

Co-author of the research study on digital transformation and knowledge worker retention in SMEs

Corresponding author on Spatial-TTT framework for streaming visual-based spatial intelligence.

Co-author of MatrAIx paper on simulated user evaluation infrastructure with 8.3 billion persona agents.

Author of "Robotics Video Benchmark and Dataset" in GitHub

Co-author of NCP-ArchPreview model development.

Researcher who contributed to SMELT paper on looped transformers cutting training compute.

Co-author of Qwen RecreationWorld paper.

Co-author of study showing endoscopists' adenoma detection rates declined after using AI tool; calls deskilling a hot research topic.

Yujia Qin

Researcher

Researcher at ByteDance

ByteDance researcher acknowledged for supporting Agent-World research and providing valuable suggestions.

Author of "Mixture-of-Recursions" on arXiv

Yujun Yan

Researcher

Researcher

Co-author of the paper on conformal prediction for temporal graphs

Yuke Zhu

Researcher

Researcher at NVIDIA

Co-authored the ASPIRE paper on robot learning framework.

Co-author with Cooney on AI-based embodied agent for art therapy

Liu was part of a research team studying how AI use affects intrinsic motivation and perceived task difficulty.

Byline author of CausalMix research on causal inference approach to LLM data mixture optimization.

Co-author of the SocioHack research from Kings College London on LLMs learning to exploit societal loopholes.

Author of "Analyzing Visual Attention in VLMs" on arXiv

Co-author and corresponding author of NCP-ArchPreview model research.

Yulin Zhang

Researcher

Computational Biology PhD Candidate at University of California, Berkeley

Co-led research team that developed TRACE method to discover archaic ancestry in modern human genomes using computational analysis.

Co-authored research on agent swarm leadership dynamics, discovering that underlying models and prompt ordering affect which agents emerge as leaders in multi-agent coordination.

Co-author on test-time scaling framework for agentic coding.

Co-author of NCP-ArchPreview research on arXiv.

Author of "800K+ Verifiable SWE Tasks" on arXiv

Yuntian Deng

Researcher

Assistant Professor at University of Waterloo

Author of "Why Can't Transformers Learn Multiplication? Reverse-Engineering Reveals Long-Ra" on arXiv

Yunze Xiao

Researcher

MS in Language Technology student at Carnegie Mellon University

Author of essay on RL training environments arguing they must encode domain unwritten rules and organizational context for agents to succeed.

Co-authored research on Agentic Rubrics for software agent verification.

Co-authored paper on SkillOpt, a text-space optimizer for agent skills at Microsoft.

Author of "Math Research Agent from DeepMind" on arXiv

Electrical and computer engineering professor leading research showing brain decision-making begins earlier than previously thought, with implications for AI architecture design.

Yushi Hu

Researcher

research scientist at Meta SuperIntelligence Labs

Author of "Training a VLM Judge Without Human Labels" on arXiv

Co-author of Mixture-of-Depths Attention research on transformer depth scaling.

Co-author of NVIDIA's JCA system for 3D animation and video co-generation.

Yuval Kansal is a researcher at Princeton University who co-authored a paper on the energy costs of AI reasoning models.

Yuval Tassa

Researcher

Researcher at Google DeepMind

Co-author of MuJoCo Menagerie, contributing to physics simulation models for AI research.

Co-authored research on visual reasoning in image generation.

Lead author of research on map-augmented geolocalization agent

Co-author of AIDE² recursive self-improvement system research.

Co-authored OpenAI Alignment post on complementary metrics for monitorability.

Co-authored research on Single-Rollout Asynchronous Optimization for training LLMs with agentic RL.

Co-author of paper on lower bounds for stepsize-based acceleration proving step-size tricks alone cannot fully speed up gradient descent.

Co-author of MoDA paper on attention mechanisms for improved transformer depth scaling.

Yuxin Zuo

Researcher

UK Education Weekly

Author of "Open-Web Simulator for Agent Training" on arXiv

Co-author of SkillLearnBench benchmark for agent skill generation evaluation.

Lead author and electrical engineering doctoral student who developed the AI model and discussed its performance metrics.

Law professor analyzing OpenAI's internal communications as potential bargaining leverage for settlement negotiations.

Wrote 'Bitter Lesson of AI CAD' post providing useful background on generalist-scaling story showing up in CAD and mechanical design.

Zachary Charles

Researcher

research scientist at Google

Author of "Tool-space interference in the MCP era: Designing for agent compatibility at sca" on arXiv

Zachary Kallenborn

Researcher

PhD candidate in Risk Analysis at King's College London

Expert discussing data centers as critical infrastructure vulnerable to physical attacks and the strategic role of AI in military operations.

Co-author of DeepMind paper on automation of alignment research

Zachary S Siegel is an author of CORE-Bench, a benchmark for AI agents on computational reproducibility tasks.

Co-author who worked on algorithm design for generating vascular networks for organ printing, achieving 100-150 micron distances from blood vessels.

Researcher at 10a Labs studying the scale and persistence of uncensored AI model redistributions.

Co-authored post on AI progress forecasts accuracy at FRI.

Noted that workers are being asked to use AI tools without adequate guidance on optimal deployment.

Zackary Nado

Researcher

Commenter on Sam Altman's Death Star messaging

Research contributor to Yale CELI's study on agentic AI adoption across industries and sectors.

Co-author of D-OPSD paper on on-policy self-distillation for step-distilled diffusion models.

Author of "Exhaustive Research Into Agentic Reasoning For LLMs" on arXiv

Co-authored GlucoFM paper on a glucose foundation model for metabolic health prediction.

Co-authored research on text-to-image generation with iterative refinement.

Researcher at Shanghai Jiao Tong University who conducted the MemoryTax study testing seven agent memory approaches.

Co-author of systematic review of code-switching speech recognition research published at LREC 2026.

Lead author of FlashDrive research paper on optimizing Vision-Language-Action model inference for autonomous driving.

Zekun Wang

Researcher

Researcher at Alibaba

Co-author on SlimQwen paper about MoE model compression.

Co-author on Microsoft research developing 4B-parameter image-to-3D model with high-resolution textured output.

Author of "Reinforcement Learning on Pretraining Data" on arXiv

Zengyi Qin

Researcher

Lead Researcher at MIT

Lead researcher on OSGym project for scalable distributed data engine for computer agents.

Co-author on Act As a Real Researcher benchmark development paper.

Co-author of MemEye research on visual-centric evaluation of multimodal agent memory.

Zewen Chi

Researcher

Researcher at Microsoft Research

Lead author of research on efficient reasoning on edge devices for language models.

Zeyi Liao

Researcher

Research Intern at NeoCognition

Author of "Meta's Agent Learning" on arXiv

Co-author on Helix World audio-visual world model research.

Listed as author of the paper on robust learning with noisy labels

Co-author on GAMUT benchmark research paper.

ByteDance AI researcher who said many Chinese AI companies rely heavily on distillation rather than building their own training data.

Co-author on the Many-Tier Instruction Hierarchy (ManyIH) research paper.

Zhang Linhao

Researcher

Researcher

Listed as author of the StableToken research paper on speech tokenization

Co-authored research on SandMLE synthetic environments for training ML engineering agents.

Listed as co-author in the CUDA Agent citation for the research on agentic RL for CUDA kernel optimization

Author of "InternVL 3.5: Open Multimodal Reasoning Model" on arXiv

Zhao Dora

Researcher

Researcher

Co-author of year-long Character.AI study examining AI companion effects on well-being and human interaction.

Foundational Models researcher at Apple who joined Anthropic

Collaborator with Xinzhi Zhang in proving the Kannan–Lovász–Simonovits conjecture using agentic AI research methods.

Zhao Song collaborated with Xinzhi Zhang to resolve a 31-year-old mathematics problem using AI-assisted research.

Zhaochong An

Researcher

Meta

Author of "Reference-to-Video Generation" in arXiv

Co-author of research paper on EvoClaw, evaluating AI agents on continuous software evolution.

Author on paper about online draft co-training for speculative decoding in RL post-training.

Core contributor on Quantum's Infinite Game project, building an AI system to discover open problems and create verifiable research environments in quantum computing.

Co-author of D-OPSD paper on on-policy self-distillation for step-distilled diffusion models.

Co-authored analysis of frontier models' performance on spatial biology tasks.

Co-author of OckBench research measuring efficiency of LLM reasoning.

Author of paper investigating how diffusion models' creativity arises from neural networks learning smoothed score functions.

Receives management advice comparing intern management to Perplexity Pro bots in extended essay on AI progress and compute.

Zhengxu Yu co-authored the Memento 3 paper demonstrating that explicit rule revision outperformed direct LLM inference on ARC-AGI-3 benchmarks.

Zhengyao Jiang

Researcher

Cofounder & CEO at Weco AI

Co-author of AIDE² recursive self-improvement system research.

Lead author of SONIC humanoid control model research published in Science Robotics.

Stanford researcher who, along with Cheng Chi and Professor Shuran Song, published a thesis on Universal Manipulation Interface (UMI) in 2024 for isomorphic robot data collection.

Co-author of Qwen RecreationWorld paper.

Co-authored research on Single-Rollout Asynchronous Optimization addressing stability and off-policy challenges in asynchronous RL.

Co-author of VIMPO paper on critic-free policy optimization for LLMs showing improvements over GRPO.

Co-author of Qwen RecreationWorld paper.

Author of "Open-Web Simulator for Agent Training" on arXiv

Contributor to Claw-Eval benchmark.

Zhihan Liu

Researcher

Lacuna

Author of "Paying Less Generalization Tax: A Cross-Domain Generalization Study of RL Traini" on arXiv

Zhihan Yang

Researcher

Private Equity Associate at Reverence

Co-author of paper on Large Language Continuous Diffusion Models (Sigma)

Author of "Anatomy of a Personal Health Agent" on arXiv

Author of "Autoregressive Flows for Image Generation" on arXiv

Zhikai Zhang

Researcher

Researcher at Tsinghua University / Galbot

Lead researcher on LATENT system that teaches humanoid robots to play tennis using imperfect human motion data and reinforcement learning.

Author of "Exhaustive Research Into Agentic Reasoning For LLMs" on arXiv

Zhipeng Cai

Researcher

Staff AI Research Scientist at Meta

Lead author of the VLM3 paper on Vision Language Models for 3D learning published on arXiv.

Co-authored research on staged image generation using planning and refinement techniques.

Zhiqiang Shen

Researcher

Researcher at Alibaba

Co-author on SlimQwen paper about MoE model compression.

Zhiting Hu

Researcher

Co-author of the paper on critiques of world models

Co-authored SkillHone framework for continual agent skill evolution with persistent decision history.

Co-authored MemT-1.0 paper on memory-native LLM architecture.

Co-authored study analyzing 8,135 trial records to reveal how AI agent skills work and fail in complex workflows.

Zhiyuan Wang co-authored the Meta Muse Glimmer technical report.

Co-author of NCP-ArchPreview latent-space model research.

Co-author of research paper on EvoClaw, evaluating AI agents on continuous software evolution.

Author of "ByteDance's Robot Learns with Language and Video" in ByteDance

Author of "WebSailor-V2: Bridging the Chasm to Proprietary Agents via Synthetic Data and Sc" on arXiv

Co-author of the SocioHack study on large language models hacking reward functions.

Researcher who worked with former doctoral student Qianzhen Shao on AI model for mapping human DNA initiator sequences.

Listed as author of the StableToken research paper on speech tokenization

Listed as author of the paper on robust learning with noisy labels

Co-author on Orchard framework paper for agentic modeling.

Co-authored the SandMLE paper on reinforcement learning for ML engineering agents.

Co-author of NCP-ArchPreview research paper.

Argued at WAIC 2026 that human-extinction narrative is part of capital-driven marketing and called out what he termed Oppenheimer-style marketing by AI companies.

Co-author of Qwen RecreationWorld paper.

Author of the ThinkingBox paper on agent benchmarking for stateful business workflows.

Co-author on the Nemotron-Cascade 2 paper with equal contribution, listed in reverse alphabetical order by first name.

Zhuoning Yuan is a researcher and co-author of VOID, Netflix's video object removal framework.

Zhuoqun Hao

Researcher

Author of "Paying Less Generalization Tax: A Cross-Domain Generalization Study of RL Traini" on arXiv

Ziang Li authored technical research on GPU-accelerated Weight Teleportation that optimizes RL training by improving weight syncing between trainers and samplers.

Co-author on the Nemotron-Cascade 2 paper with equal contribution, listed in reverse alphabetical order by first name.

Zihan Qiu

Researcher

Researcher at Alibaba

Co-author on SlimQwen paper about MoE model compression.

Zihang Dai

Researcher

Co-founder at xAI

One of nine original xAI co-founders who has departed the company in recent months.