[#VIRTUAL] PRO Workshop (AI): Scaling ML Embedding Models to Serve a Billion Queries


Senthilkumar Gopal
eBay, Senior Engineering Manager, Search ML

I am a technologist and an avid engineer focused on building solutions for complex problems and innovating with efficient architectures for applications. I have the good fortune of working on different domains such as eCommerce, Banking and Insurance and have built technology solutions for highly scalable and always available web based products, services and architecting frameworks with high fidelity for business prerogatives.

My current work involves managing multiple teams of applied researchers and engineers engaged in comprehending user provided content, specifically around unstructured item and inventory information aiding in improving information retrieval for eBay search. My work focuses on building and tuning ML models to understand and qualify item and inventory data to assist in improving search quality, creating feedback loop for sellers to understand the importance of their content and drive for clarity and conversion, and for building new experiences and other work flows in eBay.

My primary focus is on applying modern machine learning technologies and research to improve the comprehension of Item/Inventory and build semi-supervised and supervised ML models for various content related work around aspects comprehension and quality, image quality and coherence of various measures around item inventory.

I am also a conference speaker with presentations in multiple prestigious events such as API World, Silicon Valley Code Camp, ML Ops World etc., On my free time, I publish on http://www.ebaytechblog.com and also on my personal blog at https://sengopal.me. 


This talk is aimed at providing a deeper insight into the scale, challenges and solutions formulated for powering embeddings based visual search in eBay. This talk walks the audience through the model architecture, application archite for serving the users, the workflow pipelines produced for building the embeddings to be used by Cassini, eBay's search engine and the unique challenges faced during this journey. This talk provides key insights specific to embedding handling and how to scale systems to provide real time clustering based solutions for users.