# Yi Wang

> Principal Engineer at Apple

Location: Palo Alto, California, United States
Profile: https://flows.cv/yiwang1

LLM at Apple. PyTorch at Meta. SQLFlow at Ant Financial, Paddle at Baidu, Peacock at Tencent, pLDA at Google

## Work Experience
### Principal Software Engineer @ Apple
Jan 2023 – Present | Cupertino, California, United States
Apple Foundation Models

### Principal Engineer @ Facebook
Jan 2021 – Jan 2023 | Menlo Park, California, United States
- PyTorch Distributed
- Minimal Viable AI: PoC of rewriting Facebook Feed's Recommendation System using PyTorch/TorchRec.

### Principal Engineer of Deep Learning Infrastructure @ Ant Financial
Jan 2018 – Jan 2020 | San Francisco Bay Area
- Found and lead SQLFlow https://sqlflow.org, machine learning models in SQL.
- Integrate SQLFlow with OceanDB and other open-source database systems.
- Support sales to banks and insurance firms.
- https://github.com/sql-machine-learning/sqlflow

### Chief scientist and architect of PaddlePaddle @ Baidu USA
Jan 2015 – Jan 2018 | San Francisco Bay Area
- Tech lead and principal engineer of Paddle https://github.com/PaddlePaddle/Paddle
- Contributor to Deep Speech II (https://arxiv.org/abs/1512.02595)

### Lead Research Scientist @ Scaled Inference
Jan 2015 – Jan 2015 | Palo Alto, California, USA
- Lead of Research Scientists.
- Accelerate research by using Mathematica for math, cloud-based serving, and GUI.

### Senior Staff Data Scientist @ LinkedIn
Jan 2014 – Jan 2015 | Mountain View, California, USA
Built a high-order hidden Markov model of career growth patterns.

### Engineering Director of Contextual Ads @ Tencent
Jan 2010 – Jan 2014
- Design, develop and run the contextual advertising system, now known as Social Ad Platform.
- Build the team from day-one and grow it into ~100 engineres.
- Introduce code review to Tencent by enhancing Google Rietveld (https://github.com/rietveld-codereview/rietveld) to support non-Gmail login and CJK characters.
- Design and lead the development of Peacock - a latent topic modeling system that learns a million of latent topics from ad clicking data - and apply it to the Ad retrieval system. (https://dl.acm.org/doi/10.1145/2700497)

### Software Engineer @ Google
Jan 2007 – Jan 2010
- Distributed Latent Dirichlet Allocation (LDA) model in MapReduce and MPI.
- Apply LDA for product search.
- Google APAC Innovation Award (2008).

### Research Intern @ IBM
Jan 2006 – Jan 2007
Data mining and visualization at IBM China Research Lab.

### Research Intern @ Microsoft
Jan 2003 – Jan 2004
Real-time photo-realistic rendering using bi-directional texture function. Assembly programming on Radeon GPUs.


## Education
### Ph.D. in Machine Learning and Artificial Intelligence
Tsinghua University

### Research Associate in Machine Learning
City University of Hong Kong


## Contact & Social
- LinkedIn: https://linkedin.com/in/yidewang
- Portfolio: https://wangkuiyi.github.io/

---
Source: https://flows.cv/yiwang1
JSON Resume: https://flows.cv/yiwang1/resume.json
Last updated: 2026-04-12