Lei Li Email and Phone Number

To empower content interaction and creation using speech & audio related technologies.

- Lead the team to build automated SAAS products that help businesses/individuals generate illustrative/attractive content, e.g., video, audio, and text;- Deep dive into various machine-based generation techniques in the fields of NLP/CV/CG, e.g., 2D/3D virtual avatar generation, audio generation, video clip generation, natural language generation, etc.- Build efficient system to support the generation of 100k videos per day.

- Led the team to build smart call-center product with various AI and deep learning technologies, including speech recognition, text-to-speech, intent detection, response generation, etc.- Built efficient and robust call-center system to support 1.5 millions automated calls per day.

Make our search world better via machine learning techniques:- Designed and developed various machine learning models for query understanding, candidate selection and ranking, and achieved significant engagement metrics gain;- Built flexible training pipelines to train relevance models, and developed feature sharing platform across the entire relevance org.

Improve TV user experience using the latest advances of data mining and machine learning techniques.

I was a PhD. candidate at the Florida International University (FIU), working with Dr. Tao Li. My research interest involves data mining and machine learning, particularly, recommender systems. I have been working on several projects during my Ph.D. study:1. Reciprocal Recommendation: Reciprocal recommender systems refer to systems from which a list of individuals can be recommended to the target user by satisfying preferences of both parties. Different from the traditional user-item… Show more I was a PhD. candidate at the Florida International University (FIU), working with Dr. Tao Li. My research interest involves data mining and machine learning, particularly, recommender systems. I have been working on several projects during my Ph.D. study:1. Reciprocal Recommendation: Reciprocal recommender systems refer to systems from which a list of individuals can be recommended to the target user by satisfying preferences of both parties. Different from the traditional user-item recommendation, reciprocal recommenders focus on the preferences of both parties simultaneously, as well as special properties in terms of "reciprocal".2. Personalized News Recommendation: Personalized news recommender systems have gained increasing attention in recent years. With the large volume of news events happening everyday, a challenging issue of online news reading services is how to help users find interesting articles that maximally match the users' reading appetite.3. Time Sensitive Product Recommendation : In many E-commerce recommender systems, a special class of recommendation involves recommending items to users in a life cycle. For example, customers who have babies will shop on Diapers.com within a relatively long period, and purchase different products for babies within different growth stages. The goal of this project is to effectively model the diverse need of customers within different stages in a recommender system.Specifically, I(1) designed and implemented the back-end recommendation algorithms for an online recruiting system;(2) Participated in the design and implementation of FIU-Miner: An integrated and user-friendly data mining system under distributed environment;(3) Designed and implemented various recommendation algorithms for different application domains;(4) Designed and developed ontology-based multi-document summarization algorithms. Show less

I have been working on an amazing project of TV program recommendation in Digital Media Lab of Samsung Information Systems, America. This project focuses on building context-aware TV user profiles based on users’ watching histories as well as different contextual information, such as watching time, watching location, people to watch with, etc. Context-aware user profiles are generated using topic modeling approaches, and personalized TV program recommendation can be achieved by employing such… Show more I have been working on an amazing project of TV program recommendation in Digital Media Lab of Samsung Information Systems, America. This project focuses on building context-aware TV user profiles based on users’ watching histories as well as different contextual information, such as watching time, watching location, people to watch with, etc. Context-aware user profiles are generated using topic modeling approaches, and personalized TV program recommendation can be achieved by employing such profiles. Specifically, I(1) studied different hierarchical topic modeling methods on TV program synopsis data;(2) designed and developed a hierarchical topic model for context-aware user profiling;(3) designed and developed context-aware TV program recommendation algorithms. Show less

- Location-based Social CRM- Community Recommendation in Social CRM

- Topic Modeling using Wikipedia- Measuring the Effectiveness of Engagement in Social CRM