Leo Gao
Software Engineer Intern at Datadog
United States
Summary
🤩
Rockstar🎓
Top SchoolLeo Gao is a software engineer with six years of experience building ML infrastructure, data pipelines, and model evaluation tooling. He contributes to prominent EleutherAI projects—adding dataset ingestion and integrity checks for The Pile, implementing gpt2 loglikelihood and evaluation methods in the lm-evaluation-harness, refactoring attention code in GPT‑Neo with Mesh TensorFlow, and automating Kubernetes/Docker deployments and training pipelines for GPT‑NeoX. His work spans low-level model math and tokenization through to production orchestration on GPU clusters, with a strong focus on robustness and reproducibility. His GitHub bio as "Planetary Structural Integrity Engineer" hints at a systems-oriented, pragmatic mindset that prioritizes durable, auditable systems.
6 years of coding experience
6 years of employment as a software developer
Amity Regional High SchoolMathematics and Computer Science, Mathematics and Computer Science at Brandeis University
Github Skills (37)
kubernetes10
dockerce10
docker10
language-model10
data-pipelines10
python10
scripting10
data-engineering10
attention-mechanism10
machine-learning10
gpt-210
language-models10
dockers10
data-preprocessing10
tensorflow210
Programming languages (7)
Github contributions (5)
EleutherAI/lm-evaluation-harness
Aug 2020 - Dec 2022
A framework for few-shot evaluation of language models.
Role in this project:
Back-end Developer & ML Engineer
Back-end Developer & ML EngineerContributions:2 releases, 159 reviews, 516 commits in 2 years 3 months
Contributions summary:Leo primarily contributed to the development of language model evaluation methods within the framework. Their work involved adding and modifying core methods to enable model evaluation, including an `evaluate` method and implementing loglikelihood calculations for the GPT2 language model. They also integrated datasets, such as the BoolQ dataset, which highlights their focus on enabling the evaluation of models on diverse benchmarks. Furthermore, the user implemented the gpt2 loglikelihood functionality.
pytorchnlplarge-language-modelslanguage-modeldeep-learning
EleutherAI/the-pile
Sep 2020 - Jun 2021
Role in this project:Back-end Developer & Data Engineer
Back-end Developer & Data EngineerContributions:7 reviews, 162 commits, 10 PRs in 9 months
Contributions summary:Leo implemented crucial datasets and data processing pipelines for a large language model project. They added several datasets, including Deepmind Math, Enron Emails, and Literotica, expanding the scope and diversity of the training data. The user also integrated tools like checksums to ensure data integrity and included preprocessing steps and utility functions to handle and combine datasets.
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.