I build AI teams

I’m a generalist with a wide range of experiences in sales, GTM strategy, and business development across early stage startups to Amazon. 

I focus on applied AI, with a particular interest in computer vision applications. 

Outside of the above, I spend most of my time outdoors with my family (including 2 young girls + dog) and working on my pickleball game : )

Sheikh Shuvo

Member of Technical Staff

LinkedIn

1. Kernel optimization

I lead Liger-Kernel: https://github.com/linkedin/Liger-Kernel, which is the fastest growing open source project from LinkedIn ever. 

I also give a talk in CUDA MODE: https://youtu.be/gWble4FreV4?si=Ry0SPKs41KkTLba0

2. Distributed training optimization

Our optimized zero++ paper got accepted in ICML es-fomo 2024: https://arxiv.org/pdf/2407.01614

3. Inference Optimization

I am the committer in SGLang and lead the optimization around SGLang at Linkedin: https://github.com/sgl-project/sglang/commits?author=ByronHsu

Senior Software Engineer

1. ML training platform (2022/6 ~ 2022/12): 

Scale training cluster to thousands of GPUs on top of Kubernetes.

2. ML orchestration platform (2023/1 ~ 2023/5): 

Built an orchestration layer from scratch on top of Flyte by making significant improvements to Flyte, upstreaming them to open source, and also internally to serve all ML pipelines (data+training).

**Contributed to Flyte, Kuberay, Envd on the side.**

3. LLM training platform (2023/6 ~ ): 

Initiated and designed an LLM training platform on top of thousands of state-of-the-art GPUs, with performant interconnections and fast storage, enabling 100B+ scale pretraining, fine-tuning, and inference.

4. LLM distributed training (2024/1 ~ ):

Speed up internal LLM training by several times with my contribution to DeepSpeed ZeRO++.

**Contributed to DeepSpeed: https://github.com/microsoft/DeepSpeed/commits/master/?author=ByronHsu**

Software Engineer

I work on full stack of ML system:
1. Build GPU clusters on the scale of thousands of GPUs
2. Build efficient distributed and kernel GPU runtimes

I am the committer of @lmsysorg @flyteorg @theASF
You can also find me on Twitter (https://x.com/hsu_byron) or GitHub (https://github.com/ByronHsu)

taplio

The Battle

Sheikh Shuvo

Pin-Lun (Byron) Hsu

Followers

Posts

Engagements

Likes

Ready to grow your LinkedIn presence?