Ti Cheng

Senior Software Engineer (AI) at Cerence AI

Taipei, Taiwan
email-iconphone-icongithub-logolinkedin-logotwitter-logostackoverflow-logofacebook-logo
Join Prog.AI to see contacts
email-iconphone-icongithub-logolinkedin-logotwitter-logostackoverflow-logofacebook-logo
Join Prog.AI to see contacts

Summary

👤
Senior
🎓
Top School
Ti Cheng is a Senior Software Engineer specializing in LLMs and NLP with seven years of experience delivering faster, more accurate AI inference for production systems. He has hands-on expertise across Transformer architectures and inference acceleration techniques—such as speculative decoding variants (Kangaroo, Medusa, Hydra, Eagle) and ongoing Arctic Inference research—applied with tools like vLLM, SGLang, PyTorch, CUDA and Triton. Ti has led end-to-end initiatives from LoRA/QLoRA fine-tuning to building RAG systems for intelligent customer service, achieving measurable gains like 100% inference speedups and model accuracies up to 97% in commercial projects. An active open-source contributor (HuggingFace TRL, FlagEmbedding) and blogger at Clay-Technology World, he blends research-driven optimization with production engineering. Based in Taipei, he’s currently focused on fast-llm-inference and low-level GPU kernel optimization to make large models more accessible and efficient.
code7 years of coding experience
job4 years of employment as a software developer
bookMaster of Science - MS, Computer Science, Master of Science - MS, Computer Science at 國立政治大學
github-logo-circle

Github Skills (29)

pytorch10
language-model10
runtimes10
llama10
inference10
large-language-models10
tpu10
reinforcement-learning10
amd10
llm10
nvidia10
openai10
cuda10
transformer10
gpt10

Programming languages (2)

C++Python

Github contributions (5)

github-logo-circle
ccs96307/TensorRT-LLM

Jun 2025 - Jul 2025

TensorRT-LLM provides users with an easy-to-use Python API to define Large Language Models (LLMs) and support state-of-the-art optimizations to perform inference efficiently on NVIDIA GPUs. TensorRT-LLM also contains components to create Python and C++ runtimes that orchestrate the inference execution in performant way.
Contributions:40 pushes, 4 branches in 1 month
ccs96307/LeetCode

May 2021 - Jan 2023

Contributions:203 commits, 363 pushes, 1 branch in 1 year 7 months
Find and Hire Top DevelopersWe’ve analyzed the programming source code of over 60 million software developers on GitHub and scored them by 50,000 skills. Sign-up on Prog,AI to search for software developers.
Request Free Trial