avatar
Zihao Zhou
An elegant theme for Hexo

Zihao Zhou

Machine Learning PhD Student

(858)729-4258 | ziz244@ucsd.edu | La Jolla, CA

Research interests

Deep generative models, Spatiotemporal modeling

Foundation Models, Multimodality, Anomaly Detection

Education

Sep 2022 - Now (Expected: Summer 2026)

University of California, San Diego

  • PhD candidate in Computer Science
  • Advisor: Rose Yu

Sep 2020 - Mar 2022
University of California, San Diego

  • MS in Computer Science
  • GPA: 4.0
  • Advisor: Rose Yu

Aug 2016 - Jun 2020
University of California, San Diego

  • BS in Applied Math
  • BS in Cognitive Science Spec. in Machine Learning
  • Minor in Computer Science.
  • Magna Cum Laude
  • GPA: 3.941 / 4.0, Major GPA: 4.0

Publications

[1] Zhou, Z. & Yu. R., Automatic Integration for Spatiotemporal Neural Point Processes., Neural Information Processing Systems (NeurIPS), 2023

[2] Zhou, Z. & Yu, R., Automatic Integration for Fast and Interpretable Neural Point Processes., Learning for Dynamics and Control (L4DC), 2023

[3] Zhou, Z., Yang, X., He, X., Rossi, R., Zhao, H., & Yu, R., Neural Point Process for Learning Spatiotemporal Event Dynamics., Learning for Dynamics and Control (L4DC), 2022

[4] Zhou, Z., Yu, R., Can LLMs Understand Time Series Anomalies?, The International Conference on Learning Representations (ICLR), 2025

[5] Li, A.*, Zhou, Z.*, Jortberg, E. & Yu, R., Generalization of Deep Sequence Models for Forecasting Aortic Pressure Cross Cohort., Computing in Cardiology (CinC), 2022 (* equal contribution)

[6] Zhang, H., Zhou, Z., Orso, P., Pham, D., Montenegro, H., … & Caso, J. Triton RCSC 2021 Team Description Paper., Robocup Small Size League (RCSSL), 2021

[7] Sun, S., Chen, W., Zhou, Z., Jortberg, E. & Yu, R., Data-Driven Simulator for Mechanical Circulatory Support with Domain Adversarial Neural Process, Learning for Dynamics and Control (L4DC), 2024

[8] Duan, M., Qian, Y., Zhao, L., Zhou, Z., Rasheed, Z., Shafique, K., Back to Bayesics: Uncovering Human Mobility Distributions and Anomalies with an Integrated Statistical and Deep Learning Framework, 1st ACM SIGSPATIAL International Workshop on GeoSpatial Anomaly Detection (GeoAnomalies), 2024

[9] Stanford, C., Adari, S., Liao, X., He, Y., Jiang, Q., Kuai, C., Ma, J., Tung, E., Qian, Y., Zhao, L., Zhou, Z., Rasheed, Z., Shafique, K., NUMOSIM: A Synthetic Mobility Dataset with Anomaly Detection Benchmarks, 1st ACM SIGSPATIAL International Workshop on GeoSpatial Anomaly Detection (GeoAnomalies), 2024

[10] Kai, K., Tsai, Z., Zhou, Z., Yu, R., The TimeText Corpus and Multi-Modal Forecaster: a Dataset and Model for Jointly Predicting Time Series and Textual Data, arXiv, 2024

[11] Zhou, L., Yashwante, P., Fisher, M., Sampieri, A., Zhou, Z., Galasso, F., Yu, R., CaTS-Bench: Can Language Models Describe Time Series?, Findings of the Association for Computational Linguistics: ACL 2026, 2026

[12] Cao, Y., Lai, S., Huang, J., Zhang, Y., Lawrence, Z., Bhakta, R., Thomas, I.F., Cao, M., Tsai, C.-H., Zhou, Z., Zhao, Y., Liu, H., Marinoni, A., Arefiev, A., Yu, R., SimulCost: A Cost-Aware Benchmark and Toolkit for Automating Physics Simulations with LLMs, International Conference on Machine Learning (ICML), 2026

[13] (Pending) Zhou, Z., Wen, H., Yu, R., STEventsBench: A Comprehensive Benchmark for
Spatiotemporal Event Anomaly Detection

Professional Experience

Jun 2026 - Present
Co-Founder & CTO, GistFlow

• Co-founding CTO of GistFlow, an AI foresight agent platform that turns fragmented supply chain data into forecasts, research, and actionable decisions through natural language.

• Built the full-stack product from scratch, including multi-agent AI orchestration, interactive 3D globe visualization, and production infrastructure; leading engineering, hiring, and technical strategy.

Jun 2025 - Sep 2025
Apple Software Engineer Intern, Machine Learning

• Developed patent-pending Entity Schema architecture for spatiotemporal LLM agentic memory, improving personal video recall from SOTA’s 60% -> 97%

• Built dual-database system integrating graph structures with vector embeddings for real-time spatiotemporal relationship modeling across personal media assets

• Optimized spatiotemporal query latency by 90% through LLM switching and streaming architecture for Apple Photos’ on-device personalization features

Jun 2024 - Sep 2024
Machine Learning Scientist Intern, Novateur Research Solutions

• Led the development of the deep learning model for HAYSTAC project under the IARPA program, focusing on human mobility patterns.

• My work resulted in a 30% -> 66% AUPRC improvement of the deep learning model for detecting anomalies with no extra computational cost.

• Secured the highest agent accuracy score on detecting synthetic human mobility segments among five competing companies.

Jun 2019 - Aug 2019
ML Research Engineer Intern, Mininglamp Tech. Gr.

• Surveyed the current state of the art of deep temporal point process, with an emphasis on modeling dynamic knowledge graph

• Filed a patent on user identity linkage system from heterogeneous trajectories using a new trajectory similarity measure

Open Source Contributions

Gemini-OpenAI-Proxy multimodal support

• vLLM ARM build

Sakura-LLM calibration and quantization

Skills

Leading System Administrator & System Integrator

UCSD CSE Shared Servers Admin, Rose Lab Server Admin

• Hosted the lab LLM deployment frontend and backend, with lab wiki RAG

• Assessed and optimized the performance of the Nautilus Kubernetes cluster

• Authored the Nautilus User Guide

• Led end-to-end procurement and deployment for CSE shared, Roselab and SDSC share of Nautilus servers, including quote solicitation, configuration comparison, equipment purchasing, server installation, and software setup & maintenance

Software Engineer

Docker, Kubernetes, Ubuntu, Arch Linux & RHEL derivatives (CentOS, AlmaLinux), Git, PyCharm, Pytorch Lightning, Huggingface, IntelliJ, Amazon EC2, Amazon S3, MinIO

Coding

Python, Java, MatLab, C/C++, R, Go, VueJS

Media

NodeJS, Photoshop, Illustrator, Vegas, Final Cut Pro

Awards / Honors

Sep 2019 - Jul 2021
Robocup SSL Team TritonsRCSC Software Team Leader

TritonsRCSC homepage | Github codebase | Poster

· TritonRCSC is the first UCSD team that receives qualification for competing in the RoboCup Small Size League

· Designed the team software architecture (Java/C++ socket-based) and coordinated the development [4]

· Implemented the robot’s path search, obstacle avoidance, motion prediction and ball passing algorithms

Jun 2020 - Present
Phi Beta Kappa Honor Society Membership

Jun 2019 - Aug 2019
China National Smart Weather Service


Innovation Competition, 2nd Prize

Sep 2016 - Jun 2020
UCSD Provost Honor

Teaching Experience

Sep 2021 - Dec 2021
Teaching Assistant

Principles of Artificial Intelligence, CSE 250A, UCSD

Sep 2019 - Dec 2019
Instructional Assistant

Intro to Machine Learning II, COGS 118B, UCSD


Last updated: Jul 2026