# Liwei Cai

> High-frequency trading | Previously at LinkedIn, Amazon, Google | ML, NLP, C++, Python

Location: New York, New York, United States
Profile: https://flows.cv/liweicai

I am currently a Senior Software Engineer in AI at LinkedIn, improving various deep learning models and developing distributed data pipelines. I was an Applied Scientist at Amazon, working on improving Alexa's natural language understanding (NLU) ability with deep learning. Prior to that I interned at Google as a Software Engineer to make Google Translate easier to develop.

## Work Experience
### Software Engineer @ Headlands Technologies LLC
Jan 2023 – Present | New York City Metropolitan Area

### Senior Software Engineer - AI @ LinkedIn
Jan 2022 – Jan 2023 | Sunnyvale, California, United States
Designed multi-task learning neural network model for click-through rate (CTR) prediction for ads displayed on various third-party platforms, leveraging the similarity among tasks for each platform, and the larger combined training dataset.
Improved CTR prediction AUC by 1%-2%, compared to per-platform single-task production models, across all platforms in LinkedIn Audience Network, LinkedIn's third-party ads delivery system.
Developed data pipelines in Spark that processes billions of samples for production-scale model training and evaluation, model training in Tensorflow, and model evaluation in pandas.

### Applied Scientist II @ Amazon
Jan 2021 – Jan 2022 | Cambridge, MA
Led the research of end-to-end natural language understanding (NLU) deep neural network (DNN) models for Alexa, simplifying the NLU system by directly predicting user intent and keywords from audio signal. On offline test sets, it reduced error rate by 2% for all input, and by 8% for noisy input, compared to two-stage production baseline.
Launched context-aware NLU DNN models to help Alexa understand multi-turn dialog better. Built representation learning models for encoding user dialog history, and data pipelines for fetching contextual data. It reduced user dissatisfaction rate by up to 10% in online A/B tests, and was deployed into more than 5 Alexa production models.
Developed prototype DNN models in MXNet and PyTorch, run-time model serving and data pipelines in C++, and offline data pipelines for training/testing in Python and Spark.

### Applied Scientist @ Amazon
Jan 2020 – Jan 2021 | Cambridge, MA

### Teaching Assistant @ Carnegie Mellon University
Jan 2019 – Jan 2019 | Pittsburgh, PA
11-785 Introduction to Deep Learning
Developed and tested homework programs in NumPy and PyTorch.

### Software Engineer Intern @ Google
Jan 2019 – Jan 2019 | Mountain View, CA
Extended the internal debug web interface for Google Translate to enable constructing, testing and submitting new translation rules, which included developing the web frontend in HTML, JavaScript, Polymer, and the backend server in C++.
Created MapReduce-based parallel data-processing pipelines to filter test datasets from the corpus for translation rules.
Reduced the latency of adding and deploying translation rules to fix reported bad translations from hours to minutes, and enabled non-technical team members to add translation rules.

### NLP Intern (software engineering) @ Mobvoi
Jan 2018 – Jan 2018 | Beijing City, China
Designed architectures of knowledge base ontology and automatized pipeline for knowledge extraction with teammates.
Investigated and developed the knowledge storage and query module in Python and Java, which represented facts as typed RDF triples and stored them in MySQL, and translated queries into SQL with additional type check.
Migrated the existing schemaless knowledge base (stored in MongoDB) to the new type-safe one, increasing the robustness of the question answering service and the whole voice assistant ecosystem it supports.

### Research Intern @ University of California, Santa Barbara
Jan 2017 – Jan 2017 | Santa Barbara, California
Proposed a generative adversarial algorithm that can adaptively generate better negative training data according to the behavior of the knowledge graph embedding (KGE) model being trained.
Implemented the model in PyTorch and conducted experiments on various datasets to demonstrate its ability to improve the prediction accuracy of existing KGE models by up to 3%, a state-of-the-art achievement.
Authored the paper KBGAN: Adversarial Learning for Knowledge Graph Embedding as the first author and published it in NAACL 2018, a top-tier NLP conference.


## Education
### Master of Science - MS in Intelligent Information Systems (School of Computer Science)
Carnegie Mellon University
Jan 2018 – Jan 2019

### Bachelor of Engineering (B.E.) in Electronic Information Science and Technology
Tsinghua University
Jan 2014 – Jan 2018


## Contact & Social
- LinkedIn: https://linkedin.com/in/liwei-cai
- Website: https://liweicai.me

---
Source: https://flows.cv/liweicai
JSON Resume: https://flows.cv/liweicai/resume.json
Last updated: 2026-04-01