About

About Me

I am a research scientist at Biohub, where I work on generative models for biology.

Previously I was a research scientist at Springtail.ai, where I worked on sample efficient reasoning algorithms. Before that, I graduated from the University of Chicago with degrees in Computational Applied Math, Statistics, and Economics. I also worked on reinforcement learning at the Toyota Technical Institute at Chicago.

I like to share research insights on my blog and occasionally post on X.

Recent Research

• Understanding Diffusion Memorization via Complexity of the Analytical Score Target. Justin Jung. 2026. (technical report)

• Scaffold Diffusion: Sparse Multi-Category Voxel Structure Generation with Discrete Diffusion. Justin Jung. NeurIPS Workshop on Structured Probabilistic Inference and Generative Modeling, 2025. [paper] [project page]

• Guided Discrete Diffusion for Constraint Satisfaction Problems. Justin Jung. arXiv, 2025. [paper]

• Subwords as Skills: Tokenization for Sparse-Reward Reinforcement Learning. David Yunis, Justin Jung, Falcon Dai, Matthew Walter. NeurIPS, 2024. [paper]

• Scaling Test Time Inference for Protein Language Models Justin Jung. 2026. (project page)