Jiaxuan Luo

JIAXUAN LUO

Research Engineer · Streaming Multimodal Systems

jaxanluo@google.com

I'm a research engineer focused on streaming multimodal systems—real-time speech and video models, post-training, and efficient inference optimization.

Currently, I work on simulation inference at Waymo LLC, focusing on TPU/GPU inference scheduling and nondeterminism, and contribute to SGLang-Omni as a core contributor (#3). Previously, I spent three years at Alibaba developing multi-task models and RL algorithms, and conducted research with Prof. Lei Li at CMU LTI.

News

  • RASST was accepted to EMNLP 2026 Main.
  • AutoTerm-SST was accepted to EMNLP 2026 System Demonstrations.
  • Leading SGLang-Omni’s TTS runtime refactor.
  • Our vibe coding safety benchmark paper was accepted to ICML 2026.
  • Joined Waymo LLC.
  • Joined Prof. Lei Li’s group at CMU LTI.

Selected Publications

  1. CriticHack: Evaluating Visual Rewards Under Robot Policy Optimization
    Jiaxuan Luo*†, Xingguo Xu*, Shanshan Wang*, Yuhan Zhou, and Zhen Zhang
    Under review at ICLR 2027
  2. CausalCache: Conditional High-Fidelity Restoration for Long-Horizon GUI Agents
    Jiaxuan Luo, Zhanfeng Liao, Jiayao Teng, Yuan Wang, and Haojian Huang
    Under review at AAAI 2027
  3. Retrieval-Augmented Simultaneous Speech Translation
    Jiaxuan Luo, Siqi Ouyang, Jiaxing Xu, and Lei Li
    In EMNLP 2026
  4. Is Vibe Coding Safe? Benchmarking Vulnerability of Agent-Generated Code in Real-World Tasks
    Songwen Zhao, Danqing Wang, Kexun Zhang, Jiaxuan Luo, Zhuo Li, and Lei Li
    In ICML 2026
  5. AutoTerm-SST: Adaptive Terminology Memory for Simultaneous Speech Translation
    Jiaxuan Luo, Siqi Ouyang, and Lei Li
    In EMNLP 2026 System Demonstrations

Experience

2026–present

Waymo LLC

Machine Learning Engineer

Simulation inference on TPU/GPU systems, focusing on scheduling and inference nondeterminism.

2026–present

SGLang-Omni

Core Contributor

Real-time multimodal inference and shared TTS runtime systems.

2025–present

Carnegie Mellon University, LTI

Research Assistant

Streaming speech agents, retrieval, reinforcement learning, and efficient inference; advised by Prof. Lei Li.

2025

TikTok

Research Engineer Intern

Security alignment for multi-agent compliance report generation through supervised fine-tuning on internal data.

2021–2024

Alibaba Group

Algorithm Engineer (MLE)

Developed multi-task models (MMoE/MTL) and constrained RL policies for multi-objective reward optimization under budget constraints. Built causal uplift models to estimate incremental impact, achieving ~9% relative lift in targeted conversion experiments.

Selected Work

Open source · ML systems

SGLang-Omni

Core contributor (#3), leading the TTS runtime refactor and working on mixed-chunk scheduling, tensor-parallel correctness, topology-aware collectives, and shared multimodal serving infrastructure.

Research infrastructure · Speech RL

sst-rl-framework

A reusable GRPO post-training framework for speech RL with modular rollouts, rewards, typed adapters, and task-specific plug-ins.

Education

Johns Hopkins University

M.S. in Computer Science

2026

Central South University

B.S. in Computer Science · Turing Honors Program

2021

Academic Service

Reviewer: EMNLP 2026 · EACL 2027 · ICLR 2027.