top of page

Curriculum Vitae

Education

Johns Hopkins | MS (Computer Science & Robotics) 

2024 - 2025

Research Interesting: Robot Learning,  Large language model and language vision Model

Zhejiang University | MS (Software Engineering)

2014 - 2017

Thesis: “Deep Generative Model for Auto Composition by Lyrics”

Advisor: Prof. Zhang Kejun

Lanzhou University | BS (Computer Science)

2010 - 2014

Thesis: “The 3H/4W hierarchical Data Intelligence Analysis System”  (awarded as the outstanding diploma thesis)

Advisor: Prof. Zhan Jian

Education

Research
Experience

Multi-Agents Reinforcement Learning Benchmark

2022.Feb – Now

Supervisor: Assistant Prof Yang Yaodong, Peking University, Computer Science

​

Because the assortment of environments and algorithms was proposed in MARL(Multi-Agent Reinforcement Reinforcement Learning) fields recently, we want to build and propose a benchmark of current salient algorithms in different environments. We build a unified framework that could support various environments and algorithms and could run on distributed computing environment on default. In this project, I was responsible for:

 

  • Implement MAPPO, HAPPO, TRPO HATRPO, MATRPO algorithms based on Ray Rllib framework. These algorithms cloud run directly in a distributed environment.

  • Implement the unified part of these algorithms, which could run directly by configuration in different environments.

  • Fitting these algorithms to Multigent-Agent Mujoco and Start Craft SMAC environment.

 

Achievements:

  1. MAPPO, HAPPO, TRPO, and HATRPO, MATRPO could reach the original papers' proposed performance, although in our project defined framework and distributed environment.

  2. The result was proposed to ICLR 2023 and under review now. 

​

​

​

Music Generation by giving ancient Chinese Lyrics based on deep Generation Models   

2016.Sept – 2018.March

 

Supervisor: Prof Zhang Kejun, Zhejiang University, Computer Science

 

This research project explored the music generation by giving ancient Chinese lyrics. In this project, I was responsible for the following:  

 

  • Built Character Embedding for lyrics to make their semantic similarity could be calculated

  • Designed and implemented music note span embedding inspired by word embedding, which could generate a huge set of music sound elements and could calculate the acoustic similarity and coherence

  • Designed and implemented Seq2Seq with attention model to generate music elements by giving lyrics

  • The generated music sampling: https://github.com/fortyMiles/music_embedding/tree/master/dataset

Knowledge Graph and Relation Mining for Aerosol Data

2016.Jue – 2017.July

Supervisor: Prof Zhang Kejun, Zhejiang University, Computer Science

 

This research project was held by Knowledge Graph Lab, Computer Science College, Zhejiang University. Based on the papers about aerosol data and optical instrument data, get the crucial relation between the aerosol information and eight key features. In this project, I was responsible for:   

  • Implementing Tran-E algorithm to get the relation entity 

  • Implementing the Word2Vec algorithm to get the Word Semantic Similarity

  • Using Dependency Parsing to get the object-predicate-subject relation

  • Using Regular Expression and Text Parsing Method to get the table information for PDF files.

Network Society Public Opinion Distribution and Trend Mining

2013.Dec – 2014-May

Supervisor: Prof Zhan Jian, Lanzhou University, Computer Science

​

This research project held by Institute for Information, Lanzhou University and the Pennsylvania State University. Based on the Machine Learning and NLP method, mining the opinion distribution and opinion evolution trend for a given affair. In this project, I was responsible for: 

                                    

  • Implement Sentence sentimental representation

  • Implement and choose difference machine learning algorithms (Bayesian Classification, SVM, KNN, K-means) to implement the opinion classification and cluster.

  • Build graph to record the analysis the different person’s and group’s opinions;

  • Using D3 to visualize the theme river and opinion group analysis.

Achievement:  The research output paper Theme-River-Based Internet Public Opinion Visualization Correlation Analysis Methods was published on Information and Documentation Services,which is CS-SCI indexed.

Web Based Campus Service Robot Dialogue System

2012.Mar – 2013.May

Implementing an online service chatbot that could support student campus information, such as commute bus, library book information retrieval, and ordering takeout. Also, supporting the common daily conversation. In this project, I was responsible for:

 

  • Collecting corpus and standardized text data;

  • Implement the sentence semantic similarity algorithm; 

  • Implement the quick retrieval system;

  • Implement the auto-learning method;

 

Achievement:  Won the first prize in Student Research and Innovation Contest and got an outstanding award from the Ministry of Education, China. The award was the only one among more than 30 college-recommended programs to win.

Resarch Experience

Working and Project Experience

Kaikeba Technology, Beijing, Partner, Vice President, General Manager

2019.Jun – 2022.Feb

Kaikeba is an online education company with a valuation of more than one billion U.S. dollars and is the largest professional online education company in China. This company focus on Artificial Intelligence, Big Data, and Cloud Architecture. The most significant difference of Kaikeba with other similar companies is that Kaikeba focuses on advanced technology with pragmatic industrial skills, systematical training, and deeply theoretical understanding.

 

In August 2020, company received 0.55 billion RMB (80million USD) of financing led by Hillhouse Capital, which is the largest financing of online education for professional groups in China.

 

  • As the general manager of the business unit, I am responsible for marketing, operation, and human resources.

  • Responsible for constructing the Artificial Intelligence College, including curriculum system, industrial projects, learning process management

  • Responsible for the customer intelligent system. In this system, based on the data mining and machine learning methods, we could find the potential customers and satisfy customers' needs more precisely.

 

Achievements:

  1. More than 200 students I have directly taught have been admitted to artificial intelligence positions in China’s top companies, including Alibaba, Baidu, ByteDance, IBM, and Microsoft. Among the students trained by the business department, more than 1,000 people have received offers from well-known companies.

  2. 70% of the valuation of the company’s 0.55 billion RMB financing is for the business unit I am responsible for.

  3. Based on the intelligent system, we increase the turnover rate from 3% to 8%, reducing marketing costs by half.

The return of investment of my business unit is consistently higher 1:7.

IBM, Beijing, Data Scientist, China Cognition and Service Lab

2018.Mar – 2019.May

IBM Project - PCB automated manufacturing system ( 2018.August– 2019.May )

 

This project is to build for Fastprint.Inc, which is the China largest PCB(Printed Circuit Board) manufacture company located in Guangzhou.

 

In typical Chinese PBC companies, the production based on myriad requirement documents. The formats, files type, description ways all are different. The companies need 200-400 engineers to read and understand the content of the requirement documents.

 

In this project, I was the tech leader and lead data scientist.

 

  • Responsible for building the text document intelligent parser. Based on the word embedding similarity, sentence syntax and semantic parser, entity spatial relationship analysis.

  • Using computer graphic knowledge and artificial intelligence methods to parse PCB Gerber files, which are described graph files representing PCB elements and relationships. Through the parser, the system could understand the requirement of graphic representation.

  • Build the automated procedure pipeline of PCB intelligent manufacture.

 

Achievement:

  1. The related workers reduce the manual workload by nearly 70%;

  2. Finish this project, Fastprint.Inc paid 12million RMB to IBM;

  3. The project was selected as a key project in the Greater Bay Area by the Guangzhou Government

 

IBM Project - Intelligent Chatbot For China Construction Bank ( 2018.Mar – 2018. Aug )

 

This Project is to build a service chatbot to manage bank customers to complete some basic transactions. Such as information retrieval, registering a new card, close one’s bank account. In this project, I am responsible for:

 

  • Implementing the algorithm and framework for the intention classification, build models, and classify the questions into different categories, such as a daily, bank, life service, etc.

  • Using syntax tree and word2vec to generate new questions based on the customer’s questions.

  • Refactoring the existing system to make the response time faster.

 

Achievements:

  1. Complete this project successfully;

  2. The intention classification accuracy is up to 95%;

  3. Optimize the previous system whose response time is 4 – 5 seconds to less than 0.5 seconds.  

Alibaba & Ant Financial Group, Hangzhou, Algorithm Engineer (internship, Report to Ant Financial AI Director)

2013.Dec – 2014-May

Alibaba Project - Text Auto Summarization                                                      2017.Apr – 2017.Oct

 

This project uses an unsupervised learning method to implement an abstractive auto summarization system, which could convert any length article to a length shorter than 200 words article. In this project, I am responsible for:

 

  • Implement sentence embedding method to judge sentence semantic similarity.

  • Using sentence embedding to extract the main sentences.  

  • Mixed KNN algorithm to make the result more readable and fluently.

  • Using Keywords detection, NER, and dependency parsing method to get the man sentences more accurately.

 

Achievements:  

  1. Finish the project successfully, and this algorithm was used in two smart sound box which is Tmall Genie and Rokid to broadcast news.

  2. Based on the product manager’s evolution and customer feedback, the performance of this algorithm is one of the top in China, in 2017.

 

Ant Group Project - Potential Risk Works Mining                                       2017.Oct – 2018.March

 

Ant-Group is the largest Tech-Financial company. In this project, I was responsible to implement a novel weak-supervised learning method to get the risk words, based on these risk words recognize the risk and crime-related transactions in Ant-Financial (Alipay).

 

  • Design and implement the Char-Embedding, word embedding in short text transaction scenarios could catch the ‘danger similarity’ for a given word pair.

  • Using K-means and page-rank to mine more risk words.

  • Using A* search method to search more risk words based on limited known risk words.

 

Achievements:

  1. build a unified work mining system that could fit the different crime categories;

  2. increase the crucial gamble risk words from 30 to 108;

  3. find 15 criminal organizations;

  4. find the related gun transaction risk words 20, the previous number is zero.

  5. Apply a China patent “A Potential Risk Words Mining Method” successfully.

Working and Project Experience

Startup Experience

Shenzhen Beyond Distance(深圳超距科技) CEO, Founder

2018.March – 2019.March

This company aimed at training students to learn artificial intelligence systematical and pragmatically. We build a unique education path road for students who want to get offers from top Chinese companies.

 

Achievements:

  1. Our company helps over 200 students get offers from top Chinese companies, mainly positions are Natural language processing, Computer Vision.

  2. Acquired by Kaikeba, Beijing at a valuation of 30 million RMB(4.5 million USD).

Startup Experience
Publications

Honors and Awards

IBM Great Chine Group brilliance award candidates,   2019.Jan

​

Gold Award, Venture Contest, Zhejiang University,  2015.Jan

​

Outstanding Graduation Thesis, Lanzhou University,  2014.July

​

First Prize for Student Research and Innovation, Lanzhou University,  2013.May                   

​

Meritorious Winner of MCM/ICM,  USA,  2013.May 

​

Outstanding Young Volunteer, Gansu Province Government,  2012.May    

Honors and Awards

Invited Lectures

“Create a neural network framework from scratch”, Xi’an Jiaotong-liverpool University, a 5 days summer course for junior data science students. 2022.

 

“The Sequence to Sequence Theory and TensorFlow Implementation”, Python China Conference, 2017.Oct.

 

“An Overview of Natural Language Processing for Researchers”, Information Intelligence Workshop for Researchers, Lanzhou University, 2018.Feb.

Invited Lectures

Publications

Siyi Hu, Minquan Gao, Weixun Wang, Xiaodan Liang, Xiaojun Chang, Yaodong Yang, MARLlib: Extending RLlib for Multi-agent Reinforcement Learning  ICLR 2023 (Under Review)

​

Zhan Jian and Gao Minquan, Theme-River-Based Internet Public Opinion Visualization Correlation Analysis Methods, Information and Documentation Services, CS-SCI indexed, 2014, 35 (6): 17-22

bottom of page