ALAA AWAD

Metric Learning using an in-batch softmax multitask CV model with an EfficientNetB0 CNN backbone for Etsy's Search By Image product. Utilized online GPU inference, and Embedding based retrieval for visual search

Search By Image

Personalized ranking using real-time recent in-session interaction data. A dynamic user representation is generated from a custom transformer encoder, learned pretrained image representations, and on-the-fly learned embedding projection matrix. This generic personalization module is Etsy Ads approach to short-term personalization ranking and retreival models.

Personalized Ranking

Context Specific Bidding is Etsy's real time bidding system for over 100 Million Items on Etsy. Operating on 50K listings per second, the model assigns a value for every candidate ad returned on a search query.

Ad Bidding

iOS tic tac toe app with an unbeatable AI! If you want to get schooled give it a shot. Uses the minimax algorithm to create a decision tree from the fact that the two players are working towards opposite goals to make predictions

Mini Max Algorithm