Tong Gao
Research Engineer at Moonshot AI
Guangdong Province, China
Summary
🤩
Rockstar🎓
Top SchoolTong Gao is a research engineer with 9 years of experience building and maintaining production-grade ML and CV systems, currently at Moonshot AI after leading LLM evaluation and OCR efforts at Shanghai AI Laboratory. He’s a core maintainer of MMOCR and a primary developer of OpenCompass, an open LLM evaluation platform that supports models like Llama3, Mistral, and GPT-4 across 100+ datasets. His work spans full-stack development, model training infrastructure, and DevOps—contributions include SyncBatchNorm improvements for distributed training, OCR box-stitching utilities, and prompt/config alignment for large-scale evaluation. Tong combines academic rigor from UT Austin and HKUST with practical speed-ups (e.g., substantial runtime reductions during visual dialog research) and robotics control experience from RoboMaster. He is comfortable designing architecture, leading small teams, and delivering tooling that bridges research code and robust pipelines. Notably, his contributions improve both developer experience and model evaluation fidelity in widely used open-source ML tooling.
9 years of coding experience
1 year of employment as a software developer
Master of Science - MS Computer Science, Master of Science - MS Computer Science at The University of Texas at AustinExchange Computer Science, Exchange Computer Science at Washington University in St. LouisSummer Exchange Program - Computer Science, Summer Exchange Program - Computer Science at Tsinghua UniversityHong Kong University of Science and Technology (HKUST)
Chinese, Chinese, English
Github Skills (20)
pytorch10
opencv10
text-recognition10
python10
evaluation10
image-processing10
ocr10
machine-learning10
cicd10
llm10
ocra10
text-detection10
sd-mmc10
computer-vision10
benchmark9
Programming languages (3)
Github contributions (5)
open-mmlab/mmocr
Jun 2021 - Jan 2023
OpenMMLab Text Detection, Recognition and Understanding Toolbox
Role in this project:
Backend & DevOps Engineer
Backend & DevOps EngineerContributions:16 releases, 848 reviews, 360 commits in 1 year 7 months
Contributions summary:Tong's contributions focused on implementing OCR box stitching functionality within the mmocr library. They added a utility to stitch fragmented word boxes into lines, incorporating functions for vertical overlap checking, calculating distances, and merging boxes. Furthermore, the user updated the demo application, including code to group OCR results into lines. The user also added support for a TextOCR dataset converter to aid in text recognition. Additionally, the user made substantial improvements to the CI/CD pipeline by including CI.
panettextsnakeopenmmlabcrnnunderstanding
open-compass/opencompass
Jul 2023 - Oct 2023
OpenCompass is an LLM evaluation platform, supporting a wide range of models (Llama3, Mistral, InternLM2,GPT-4,LLaMa2, Qwen,GLM, Claude, etc) over 100+ datasets.
Role in this project:Full-stack Developer
Full-stack DeveloperContributions:4 releases, 237 reviews, 179 PRs in 3 months
Contributions summary:Tong primarily focused on aligning and updating prompt files, which suggests a role in configuring and managing prompts for the LLM evaluation platform. They implemented fixes and made minor modifications to the code in various configuration files related to datasets and evaluation. The user also translated lark messages and enhanced the `run.py` script, indicating contributions to the project's core functionality and user experience.
benchmarkchatgptevaluationlarge-language-modelllama2
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.