Lianmin Zheng

Lianmin Zheng builds AI systems that span deep-learning compilers, distributed training, LLM inference, and model evaluation. He coauthored TVM in 2018 and Alpa in 2022, then helped introduce vLLM, Chatbot Arena, LLM-as-a-Judge, and SGLang in 2023. He completed a PhD at UC Berkeley's EECS department in 2024 and holds a B.S. from Shanghai Jiao Tong University's ACM Honored Class. In October 2025, he led xAI's inference team for Grok models; as of October 2026, he is an engineer at Meta.

Now at: Meta
Badges: Bridge
Links: Website, Google Scholar, X, LinkedIn and GitHub

Timeline

2018

TVM across hardware

In 2018, he coauthors TVM, an open-source compiler that uses graph-level and operator-level optimizations to make deep-learning workloads portable across diverse hardware backends.
2022

Alpa for parallel training

In 2022, he coauthors Alpa, which automates model-parallel training of large deep-learning models through execution plans that unify data, operator, and pipeline parallelism.
June 2023

Chatbot Arena and LLM judges

In June 2023, he coauthors “Judging LLM-as-a-Judge with MT-Bench and Chatbot Arena,” a study that introduces MT-bench and Chatbot Arena to compare LLM judges with human preferences.
June–September 2023

PagedAttention and vLLM

In June 2023, the team introduces vLLM, an open-source library for fast LLM inference and serving. Zheng coauthors its September 2023 PagedAttention paper, which reports larger throughput gains for longer sequences and larger models.
December 2023

SGLang for structured programs

In December 2023, he coauthors SGLang, a system for efficient execution of structured language-model programs. Its frontend supports generation and parallelism controls; RadixAttention reuses the KV cache, and compressed finite-state machines speed structured-output decoding.
2024

A Berkeley PhD

In 2024, he completes a PhD through UC Berkeley's EECS department under advisors Ion Stoica and Joseph Gonzalez. His dissertation covers Alpa for model-parallel training, SGLang for inference, and Chatbot Arena and LLM-as-a-Judge for evaluation. He also holds a B.S. from Shanghai Jiao Tong University's ACM Honored Class.
October 2025

Grok inference at xAI

In October 2025, he is a member of technical staff at xAI, where he leads the inference team responsible for efficient and reliable infrastructure for Grok models.
As of October 2026

Engineer at Meta

As of October 2026, he is an engineer at Meta.