San Jose, California, United States
I am passionate about understanding, extending, and applying the capabilities of LLMs and other forms of Predictive, Generative AI and Agents. Specialties: deep learning, reinforcement learning, machine learning, algorithms, computational advertising, recommender system, information retrieval, distributed system, Amazon web service
- designed and developed an autonomous financial agent system in training and inference, which incorporates machine learning techniques and LLMs.
Advertising Science predictive ads bidder * problem definition, metrics definition, experiment design, online/offline data analysis and visualization * feature engineering, features selections, transformations and model selections * built end-to-end distributed trainers, customized objective functions and tune hyper parameters * online model deployment and inference Advertising Exchange Platform * built advertising experimentation framework - online A/B experiments * built advertising exchange platform - low latency(<90ms), high volume(>Mqps) * built advertising identity system across multiple regions with NoSQL
* collaborated with scientists to apply predictive models for targeting/scoring framework in Ads Servers * designed and developed scalable User Profile Servers (Dynamo like) for geo/demographic/behavioral targeting between data centers from both online and offline clustered HDFS replicator data through message bus (300K QPS, 10ms)
Search and Data Analytics Cloud Service * lead team(10) to design and build multi-shards, fault-tolerant search farms with Lucene for searching user profiles, document and meeting minutes
Core and Platform Group * One-to-One personalization servers ( Highly scalable, C++ multi-threaded, JNI, Embedded JVM, Java multi-threaded) * Load Balancers (C++), Cache Framework, Clustered Servers, SaaS
One-to-One Personalization Servers