Szi-chieh Yu
Data and AI Enthusiast
About
Hi 👋, I'm Szi-chieh Yu, a program manager on the human data team at OpenAI, where I help build the data collection pipelines behind powerful AI models.
Before entering tech, I spent over a decade as a neuroscience researcher. My work focused on synaptic recycling and whole-brain connectomics — from mapping the Drosophila brain to tracing mouse visual cortex circuits, including the first complete connectome of an adult fly brain and the largest connectome dataset. I’ve worked across research institutes in the US and Europe, including Princeton, Goethe University, and UIC, and collaborated with researchers around the world!
At OpenAI, I’ve led data initiatives across multimodal and reasoning model development — from image generation to mathematical reasoning, including the efforts that brought our model to IMO gold-level performance. I’m particularly interested in the intersection of people, data, and AI — and how human insight can shape machine learning in meaningful ways, turning complex challenges into possibilities.
Outside of work, you’ll usually find me hunting for good stationery, tinkering with workflows, hiking, traveling, exploring international cuisines, or recharging with coffee and books.Get in touch!
numbers
Work
At OpenAI
Data | GPT-5.2 Slide GenerationData & Evaluation | GPT Image Generation 1.5Data | GPT-5’s multimodal capabilities with more accurate visual reasoningGold medal-level performance 🥇on the 2025 International Mathematical Olympiad
Research Contributor | GPT-4.1 in the API
Contributor (Data) | GPT-4o Image Generation
At Princeton University
The MICrONS Project
The "Apollo Project of the Brain" - Obama’s BRAIN Initiative, a $100 million effort to decode brain function and inspire algorithms that bring human-like reasoning to AI system. It's the largest wiring diagram and functional map of the brain to date. (2025)
My contribution highlights:
1. Functional connectomics spanning multiple areas of mouse visual cortex
2. Functional connectomics reveals general wiring rule in mouse visual cortex
3. Perisomatic ultrastructure efficiently classifies cells in mouse cortex
4. Connectomics of predicted Sst transcriptomic types in mouse visual cortex
5. NEURD offers automated proofreading and feature extraction for connectomics
6. Reconstruction of neocortex: Organelles, compartments, cells, circuits, and activity
and more...
The FlyWire Project
The FlyWire Connectome is the first complete synaptic-resolution map of an adult fruit fly (Drosophila melanogaster) brain, made possible through AI-driven methods and global crowdsourced collaboration. Its success has provided groundbreaking insights into neural circuitry, significantly advancing our understanding of how brain structure supports function and behavior.
My contribution highlights:
1. Neuronal wiring diagram of an adult brain
2. Neuronal parts list and wiring diagram for a visual system
3. Whole-brain annotation and multi-connectome cell typing of Drosophila
4. Network statistics of the whole-brain connectome of Drosophila
5. The fly connectome reveals a path to the effectome
6. A Drosophila computational brain model reveals sensorimotor processing
7. Neurotransmitter classification from electron microscopy images at synaptic sites in Drosophila melanogaster
and more...
Invited Talks (selected)
Invited Speaker | WiDS Taipei Conference 2025
Presenter | MICrONS Technical Exchange Meeting, IARPA, 2020
Presenter | MICrONS kickoff Meeting, IARPA, 2019
Presenter | C. elegans Topic Meeting, Japan, 2016
Presenter | Channelrhodopsin et al. - Optogenetic tools and applications, Germany, 2014
Speaker | Night of Science, Goethe University, Germany, 2014
Speaker | Goethe University, Buchmann Institute for Molecular Life Sciences, Germany, 2012
Presenter | 5th East Asia Worm Meeting, Taiwan, 2012
Presenter | EMBO conference series - C. elegans Neurobiology, Germany, 2012
Contact
Feel free to reach out! I’m always open to collaboration, new ideas, and "useless" chats over coffee!