Faster recommendations with user-item features: MySQL HeatWave's two-tower model
Blog post on the two-tower model in MySQL HeatWave AutoML, leveraging user and item features for faster, more accurate recommendations.
Currently I work within the MySQL AI X team, deploying Generative AI solutions across the MySQL service. There, I built the integration of LLMs (OpenAI, Meta, Cohere) with the MySQL Database Service. Before that, I led Large Language Model and Generative AI deployment in the MySQL HeatWave cloud service, covering in-database LLM inference, RAG, vector search, and HeatWave AutoML recommendation systems. Selected talks and write-ups below; full timeline in the CV.