Joshua Hong
Software Engineer at Google
Pittsburgh, Pennsylvania, United States
Summary
🤩
Rockstar🎓
Top SchoolJoshua Hong is a software engineer specializing in scalable distributed systems, ML compilers, and cloud infrastructure with five years of experience and current roles at Google and Carnegie Mellon University. He has production experience building high-throughput inference and content-moderation pipelines that process millions of daily requests and has modernized ranking and streaming pipelines at major tech companies. An active open-source ML engineer, Joshua has contributed quantized operator support to the widely used TVM compiler and helped integrate many LLM architectures into the MLC-LLM deployment engine, improving performance and KV cache behavior. His research background spans LLM reasoning and multilingual word alignment, with published work on planning-style reasoning with language models. Based in Pittsburgh, he combines rigorous academic training (MS CMU, BS UCSD) with hands-on systems engineering, often tackling low-level numeric formats (bfloat16/quantization) to squeeze performance out of real-world ML systems.
5 years of coding experience
3 years of employment as a software developer
University of California, San DiegoComputer Science, Computer Science at The University of OsakaHigh School Diploma, High School Diploma at Monta Vista High SchoolMaster of Science - MS Computer Science, Master of Science - MS Computer Science at Carnegie Mellon University
English, Chinese
Stackoverflow
111reputation
19kreached
1answer
1question
Github Skills (20)
integrate10
python10
machine-learning10
integrations10
deeplearning-ai10
compiler-compiler10
deep-learning10
compiler10
quantization10
tflite10
data-integration10
testing9
tvm9
tensor9
curl6
Programming languages (3)
Github contributions (5)
mlc-ai/mlc-llm
Oct 2023 - Dec 2024
Universal LLM Deployment Engine with ML Compilation
Role in this project:
ML Engineer
ML EngineerContributions:4 reviews, 31 PRs, 2 branches in 1 year 1 month
Contributions summary:Joshua primarily contributed to adding support for new architectures within the ML deployment engine. Their work focused on integrating the Baichuan2, InternLM, StableLM, Qwen2, ChatGLM3, InternLM2, Starcoder2, MiniCPM, Deepseek, and GPTJ architectures into the system. This involved updating model-related files, modifying conversation templates, and migrating models to PagedKVCache for performance improvements.
language-modelllmmachine-learning-compilationtvm
apache/tvm
Aug 2023 - Apr 2025
Open deep learning compiler stack for cpu, gpu and specialized accelerators
Role in this project:ML Engineer
ML EngineerContributions:7 reviews, 58 PRs, 38 comments in 1 year 8 months
Contributions summary:Joshua's commits primarily focus on implementing and supporting quantized operations within the TFLite frontend of the TVM compiler stack. They added support for several quantized comparison and arithmetic operators such as LESS, NOT_EQUAL, GREATER_EQUAL, LESS_EQUAL, and POW. They also added support for ELU quantization in the TFLite frontend. The contributions involve modifying TFLite frontend code and adding corresponding tests.
metalvulkancompilertensoropencl
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.