Jonathan Chang - Personal Website

Jonathan Chang, machine learning engineer with focus on LLMs and generative models

Jonathan Chang

I'm a machine learning engineer focused on LLM inference and AI research. I build open-source tools and experiments.

email: contact at jonathanc dot net

Chart comparing mixture-of-experts maximum load violation across causal routing methods

Causal Routing Bias for Aux-Loss-Free MoE Training

Research, 2026

Causal router-side balancing methods for aux-loss-free MoE training

Performance comparison chart showing tokens per second between flex-nano-vllm and vLLM

flex-nano-vllm

Code, 2025

A minimal vLLM inference engine built from scratch using PyTorch FlexAttention

MIST robot prototype showing expressive face

Project MIST

Experiment, 2025

Building the MIST robot prototype with AI-assisted design

Experiments, benchmarks, and writing made by AI agents, with my direction and feedback.

A simulated head and throat singing, beside the title Rickrolled by a simulated throat
Agent-made · Experiment

Rickrolled by a simulated throat

Claude Opus 5.5 built a voice whose simulated vocal folds vibrate on their own, and fitted it to Rick Astley: likely the first whole song sung this way.

Built and written by Claude Opus 5.5, with direction and listening feedback from Jonathan Chang.

A green target swatch beside predictions from GPT-6.2 Astra and Claude Haiku 5.5
Agent-made · Benchmark

Color Accuracy Bench

Can vision models estimate a solid color as a hex code? Compare their answers on 100 sRGB swatches.

Made with GPT-6.2-Sol, with direction from Jonathan Chang.

Can AI read this? Six tiles spell 你看得懂嗎? — Can you read it?
Agent-made · Benchmark

Traditional Chinese Bench

How vision models read 1,200 isolated Traditional Chinese characters, from common text to rare forms.

Made with Codex, with direction and feedback from Jonathan Chang.

Local fine-tuning through selective neurons
Agent-made · Experiment

Local fine-tuning through selective neurons

Train only the output weights of a few neurons that fire in the target context; the model then changes 3-19× less on unrelated text than with a LoRA.

Written and run by Claude Opus 5.5, with direction and feedback from Jonathan Chang.

Examples of one neuron steering a chat model toward different personas
Agent-made · Tool

neuron-finder

Find individual MLP neurons that steer a chat model toward a named persona, with ranked candidates and results for 46 targets.

Created by Claude Opus 5.5 and Jonathan Chang.

A color-dot plate containing the number 23
Agent-made · Benchmark

Colorblind Bench: Can LLMs read a color vision test?

Colorblind Bench tests whether models can read numbers defined by color, using 140 color plates and 120 controls.

Made with Codex and Claude Code, with direction and feedback from Jonathan Chang.

Blog

Timeline

2026 - present · NVIDIA

I work on inference in the Nemotron team.

2025 - 2026 · Poolside AI

I worked in the inference team, where I helped optimize inference and worked on model quantization.

2022 - 2025 · Taboola

Spend some time working in algorithm team, worked on feature engineering and designed experiments. Later joined the Generative AI team, where I integrated and optimized SoTA image models into our product.

2021-2022 · BigScience Project

I contributed to the BigScience project, mainly in the metadata working group. I worked on the training codebase and conducted research experiments on using metadata to improve language model performance.

2021-2022 · ASUS AICS

I spent 6 months in the AICS department, where we collaborated with local hospitals. I worked on medication recommendation models using BERT.

2020-2021 · NTU MiuLab

I worked with Prof. Yun-Nung Chen on SoTA Dialogue System based on GPT2. We participated in the Situated Interactive MultiModal Conversations (SIMMC) Challenge (DSTC9) and achieved 3rd place.

2020 · Google

Spent a summer at Google, internship was replaced with a remote project due to COVID-19. I worked on a project using NLP to recommend relevant articles to users.

2017-2021 · National Taiwan University

Completed my undergraduate studies in Computer Science, focusing on machine learning and artificial intelligence. I was a TA for the course Applied Deep Learning.

Frequently Asked Questions

What is LLMProc and how does it differ from traditional agent frameworks?

LLMProc reframes LLM applications as Unix-like processes rather than agents. It provides powerful abstractions like fork() for parallel exploration, goto() for context management and time-travel debugging, and a file descriptor system for handling large tool outputs. This process-based approach makes scaling and managing LLM execution more intuitive using familiar computing paradigms.

How does minLoRA achieve such parameter efficiency?

minLoRA implements Low-Rank Adaptation (LoRA) in just ~100 lines of code. It freezes pre-trained model weights and injects trainable low-rank decomposition matrices, dramatically reducing trainable parameters from millions to thousands. Built on PyTorch's native parametrization system, it works seamlessly with any torch.nn.Module without modifying model definitions.

What makes Additive Rotary Embedding different from standard RoPE?

Additive Rotary Embedding (AddRoPE) modifies RoPE by making position encoding additive rather than multiplicative, with learnable weights and phase offsets. This allows models to selectively ignore certain frequencies and provides more natural attention patterns. In experiments, it matches or slightly outperforms RoPE while being computationally faster.

What programming languages and frameworks do you primarily work with?

I primarily work with Python for machine learning projects, using frameworks like PyTorch, Transformers, and FastAPI. For web applications, I use modern JavaScript/TypeScript with tools like Bun and Tailwind CSS. I also have experience with shell scripting and Unix-based systems, which influences my approach to building tools like LLMProc.

Are your projects open source and how can I contribute?

Yes, all my projects are open source and available on GitHub. I welcome contributions! Each repository has its own contribution guidelines, but generally I appreciate bug reports, feature suggestions, documentation improvements, and code contributions. Feel free to open issues or pull requests on any project that interests you.