Chengxi Li
Member of Technical Staff, Fireworks AI — post-training infrastructure & agentic RL for LLMs.
BasedBay Area
LinkedInchengxi-li GitHub@Hecate0821
EducationCMU M.S. · Xidian B.Eng.
Chengxi Li
Experience
2026 — Present

Fireworks AI — Member of Technical Staff

Post-training infrastructure and agentic RL for the Fireworks LLM platform.
Fireworks AI · Blog
Scaling and Optimizing Frontier Model Training
Fireworks Training SDK — the model catalog, parallelism stack, precision kernels, and memory optimizations for fine-tuning trillion-parameter MoE models.
→
San Mateo
2025 · Jun — Dec

SGLang RL Team — Contributor (slime · SGLang)

Open-source contributor for RL systems.
Remote
Blog
2026 · May

The Load-Balance Problem Behind Hybrid Parallelism

A compact note on Megatron Dynamic-CP, ByteScale Hybrid Data Parallelism, and why DP, CP, and PP become coupled when sequence lengths vary.
DP · CP · PP
Focus
LLM Post-Training Agentic RL