The Fifth Elephant 2023 Monsoon

The talk will be an implementation story which falls under #5 in the Topic for Submission - [https://hasgeek.com/fifthelephant/2023/sub#h:topics-for-submission] section. It will be mainly focused on a problem statement in the Search domain to rank the search results retrieved from the Search engine. It explains in detail about the advanced/SOTA solutions using Deep Learning, Reinforcement Learning, and deploying those heavy large models in the cloud with cutting-edge technologies contrary to the traditional Learning to Rank approaches. It also talks about several models/approaches that can be taken in the journey of experimentation stages. Since the talk is based on implementation story it will provide a lot of practical insights to the audience, especially in Ranking and Recommender systems.
The key takeaway for this talk will be learning about Deep Reinforcement Learning (Point/Pair/Listwise Loss) with the Transformers-based model in PyTorch, Feature engineering, and their production deployment for high throughput and low latency serving on Container in AWS Cloud.