Theory, engineering, and deployed systems. 理论、工程与可部署系统。

I trained as a mathematician at Virginia Tech and started professionally in sonar, signal processing, and autonomy. That combination has stayed with me: I like formal ideas, but I am most interested when they have to survive a real system.

我在 Virginia Tech 接受数学训练，职业早期做声纳、信号处理与自主系统。这种组合一直影响我： 我喜欢形式化思想，但最感兴趣的是这些思想能否在真实系统中成立。

At Metron, I worked on U.S. Navy sonar and underwater vehicle systems, including machine-learning and signal-processing work for the Knifefish unmanned undersea vehicle. It was a useful education in what happens when data, physics, operators, and deployment constraints all matter at the same time.

在 Metron 时，我参与美国海军声纳与水下无人系统工作，包括 Knifefish 无人水下航行器的机器学习与信号处理。 那段经历让我很早就理解：数据、物理环境、操作者和部署限制往往必须同时考虑。

At Lawrence Livermore National Laboratory, I built and led applied machine-learning and statistics groups, served as a principal researcher across more than $20M in research, and worked on systems involving text, images, video, audio, communications, graphs, recommendations, retrieval, uncertainty, and generative models.

在 Lawrence Livermore National Laboratory，我建立并领导应用机器学习与统计团队，并作为主要研究人员参与超过 2000 万美元的研究项目。工作覆盖文本、图像、视频、音频、通信数据、图、推荐、检索、不确定性与生成模型。

I am now a Senior Research Fellow at Zhejiang Lab, working on GeoGPT and AI systems for scientific research; a Visiting Professor at Zhejiang University; and CTO of Imbrial AI, where the interest is custom AI that learns the user's context rather than treating everyone as the same prompt.

我现在是浙江实验室高级研究专家，从事 GeoGPT 与面向科学研究的 AI 系统；同时担任浙江大学访问教授与 Imbrial AI CTO，关注能够学习用户真实上下文的定制 AI，而不是把所有人都当成同一个提示词。