AI researcher partner. 懂研究,更懂研究者。

TensorAI connects frontier AI labs with researchers who can shape what comes next. We look beyond author lists—to hypotheses, experiments, ablations, and code—to understand what a team needs, what a researcher wants to pursue, and where the two genuinely align. Confidential by default. No researcher is identified or presented without permission. TensorAI 连接前沿 AI 实验室与能够定义下一步的研究者。我们不止看作者名单,也会深入理解研究假设、实验设计、消融实验和代码,判断团队真正需要什么、研究者下一步想探索什么,以及双方是否真正契合。默认保密;未经本人同意,不透露身份,也不对外推荐。

The human tensor研究者的多维信号

A paper is a signal.
The work behind it is the evidence.
论文只是信号,
背后的工作才是证据。

01

Contribution, reconstructed还原真实贡献

Across papers, repositories, talks and follow-up work, we ask who framed the question, built the system, designed the decisive ablations and resolved the failures.我们沿着论文、代码仓库、公开分享与后续工作,追问是谁提出问题、搭建系统、设计关键消融实验,又是谁解决了让研究受阻的难题。

02

Originality & trajectory原创性与研究轨迹

We distinguish excellent execution from the ability to set a new research agenda—through problem selection, hypothesis quality, experimental judgement, and responses to negative results—not just venues, citations, or affiliations.我们区分沿既有方向的出色执行与开辟新研究议程的能力,考察问题选择、假设质量、实验判断以及如何应对负结果,而不只看发表会议、引用量或机构背景。

03

The research multiplier研究者的乘数效应

The strongest researchers set agendas, create reusable methods and evaluations, raise scientific standards, mentor others, and attract other exceptional researchers. We assess how that impact compounds across an organisation.最出色的研究者能够设定研究议程、沉淀可复用的方法与评测体系、提升科研标准、培养他人,并吸引更多优秀人才。我们评估的,正是这种对组织的放大效应。

Research-defining hires定义研究方向的关键研究者

The right researcher
for the right programme.
让对的研究者
遇见对的研究计划。

01

Research leaders & principal scientists研究负责人及资深科学家

Chief scientists, VPs and heads of research, principal scientists, and research leads who can identify consequential problems, set the scientific bar, and turn insight into a durable research programme.面向首席科学家、研究副总裁或研究负责人、资深科学家及方向负责人——他们能够选择真正重要的问题、设定科研标准,并将洞见发展为可持续的研究计划。

02

Research scientists & engineers研究科学家与研究工程师

Targeted searches across papers, code, benchmarks, labs, and open-source communities—grounded in an evidence-based view of each researcher’s contribution, originality, and future potential.我们跨论文、代码、基准评测、实验室网络与开源社区开展定向寻访,并基于证据判断候选人的真实贡献、研究原创性与未来潜力。

03

Research team build-outs研究团队组建

We translate a research roadmap into a capability map spanning algorithms, data, evaluation, and systems. We then sequence the hires and define how the roles work together.我们将研究路线图拆解为算法、数据、评测和系统能力图谱,确定关键人才的引进顺序,并设计让不同能力高效协同的团队结构。

04

Frontier talent maps前沿研究者图谱

Field maps built at the level of individual researchers: established leaders, fast-rising researchers, and overlooked contributors whose work carries more weight than their visibility suggests.构建深入到作者层面的领域人才图谱,识别公认的领军者、快速成长的新锐研究者,以及贡献尚未被知名度充分反映的人才。

AI research frontiers算法前沿

Focused on the fields shaping frontier AI.聚焦推动前沿 AI 演进的核心研究领域。

Foundation Models & Pre-training基础模型与预训练Architectures, scaling laws, data mixtures, optimisation and distributed pre-training模型架构、规模定律、数据配比、优化算法与分布式预训练
Post-training, RL & Reasoning后训练、强化学习与推理SFT, reinforcement learning, reward modelling, long-horizon reasoning and test-time computeSFT、强化学习、奖励建模、长程推理与测试时计算
Agents, Environments & Tool Use智能体、环境与工具使用Agents for coding and computer use, environments, memory, planning, and multi-agent learning工具调用、代码与计算机操作智能体、环境设计、记忆、规划与多智能体学习
Multimodal & Generative Models多模态与生成模型Vision-language models; image, video, and audio generation; and unified multimodal understanding视觉语言、图像、视频与音频生成,以及统一多模态理解
World Models & Embodied AI世界模型与具身智能Learned simulators, spatial intelligence, perception–action learning, control and robotics学习型模拟器、空间智能、感知—行动学习、控制与机器人学习
Alignment, Evals & Safety对齐、评测与安全Interpretability, monitorability, robustness, frontier evaluations, oversight and alignment可解释性、可监测性、鲁棒性、前沿能力评测、监督机制与对齐

Contact us联系我们

For the labs building frontier AI
and the researchers who make it possible.
与前沿 AI 实验室同行,
也与研究者共创突破。

Our long view · The right connections between frontier labs and exceptional researchers can accelerate scientific progress—and help bring AGI closer.长期愿景 ·让合适的实验室与研究者彼此成就,让科研进步产生复利,为人类进步与通往 AGI 的探索贡献一小步。