Gavin Li
Founder at animaai.cloud
San Francisco, California, United States
Summary
🤩
RockstarGavin Li is a founder and engineer based in San Francisco with 13 years of experience building back-end systems and production ML tooling. As CEO of Anima AI, he blends hands-on model engineering with product leadership, shipping practical LLM fine-tuning workflows like QLoRA for Chinese-language models. His open-source contributions span high-performance distributed computing and ML infra—from enhancing Spark’s pipe mechanism and adding snappy compression to enabling 70B-model inference on a single 4GB GPU. Comfortable across Python, PyTorch, Transformers and core backend stacks, he focuses on making large models and cluster workflows more efficient and accessible. Gavin’s work shows an uncommon mix of systems-level optimization and ML-centric experimentation, reflecting both startup urgency and deep engineering rigor.
12 years of coding experience
Github Skills (14)
scala10
transformers10
javas10
pytorch10
apache-spark10
python10
fine-tuning10
java10
data-processing10
distributed-computing10
llm10
testing9
lora9
compression7
Programming languages (7)
Github contributions (5)
lyogavin/airllm
Jun 2023 - Jan 2025
AirLLM 70B inference with single 4GB GPU
Role in this project:
Back-end Developer & ML Engineer
Back-end Developer & ML EngineerContributions:10 PRs, 191 pushes, 1 branch in 1 year 7 months
Contributions summary:Gavin's contributions primarily involve developing and refining a QLoRA training script for a large language model (LLM). They implemented core training functionality using Python and libraries such as PyTorch, Transformers, and PEFT. The user updated the training script, including dataset integration for Chinese-based LLMs and evaluation metrics, which demonstrates a focus on model finetuning and related optimization.
chinese-llmchinese-nlpfinetunegenerative-aiinstruct-gpt
mesos/spark
Jun 2013 - Jul 2013
Lightning-fast cluster computing in Java, Scala and Python.
Role in this project:Back-end Developer
Back-end DeveloperContributions:10 commits in 1 month
Contributions summary:Gavin enhanced the Spark `pipe` functionality, a feature for executing external processes within Spark RDDs. Their work includes adding new options for transforming data passed to the external process and providing a mechanism to include context data, improving flexibility. The user added unit tests to validate the new features and addressed comments, indicating an iterative development approach. They also introduced a compression codec and snappy compression support to optimize data storage.
pythoncluster-computinglightningsparkscala
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.