San Francisco, California, United States
I am a Staff Software Engineer at MongoDB, working on vector search in Atlas Search. Previously, I was part of Visa’s GenAI Platform team, where I built an LLM platform to efficiently serve LLM queries. Before that, I worked on Visa’s Machine Learning Platform team, developing an end-to-end machine learning solution covering feature engineering, distributed training, and automated deployment. Earlier in my career, I was on the Amazon Redshift team, applying machine learning techniques to enhance database performance through intelligent resource allocation, admission control, and starvation control. This work led to a publication in SIGMOD'23 (https://dl.acm.org/doi/10.1145/3555041.3589677) During my PhD, my research focused on: - Building high-performance database systems - Building efficient search systems More about me: http://chunbinlin.com
Vector Search - working on vector search in Atlas Search (https://www.mongodb.com/products/platform/atlas-search)
- Genai platform: Leading the team to build a platform for model serving. - Machine learning platform: Leading the team to building high performance machine learning platform for feature generation and model deployment.
- Automatic workload management by applying machine learning models. - Data sharing query processing - Concurrency scaling - Result cache - Query optimization by skipping data - Query monitoring rules - Machine learning model trainer
- GQFast: In-memory column database. GQFast supports fast SPJA SQL queries over compressed data. - Plato: Time series database. Plato provides deterministic error guarantees for analytic queries over compressed time series.
Big Nested Data Management - Propose SQL query rewritting algoritms to rewrite SQL queries with nested information