Currently, I work as a data scientist at Wolt.

Prior to this, I worked as a senior data scientist at Delivery Hero in ranking and personalization team. I collaborated on a machine learning-driven cross-sell project, emphasizing product ranking. I conducted extensive exploratory data analysis on cross-sell events and developed models for recommending complementary products. I oversaw end-to-end deployment of these models and designed offline metrics that correlated with online performance indicators. Additionally, I effectively managed stakeholder relationships throughout the project.

Previously, I worked as a senior data scientist at JustEat Takeaway on large scale search and recommender systems. I have been focussing on both research and engineering aspects of large scale information retrieval systems, using state of the art techniques. I have also been working on removing various sorts of biases from ranking lists. I have built and productionalized implicit matrix factorisation for personalised restaurant recommendations using SageMaker, which has been used to replace popularity based baseline in personalized restaurant emails. Additionally, I have built a POC for personalised restaurant recommendations using Factorisation Machines. In addition to this, I worked on unbiased learning to rank by building a position-bias corrected click-through-rate using search logs. Finally, I worked on counterfactual off-policy evaluation (OPE) of general purpose ranking lists.

I obtained my Ph.D. in recommender sytems in online advertising from University Grenoble Alpes, a prestigious university in Grenoble, France. My Ph.D. thesis was supervised by Prof. Massih-Reza Amini and co-supervised by Assoc Prof. Charlotte Laclau, on recommender systems in online advertising.

You can find more information in my resume Sumit Sidana

Research Interests

Recommender Systems, (Unbiased) Learning to Rank, Contextual Multi-armed bandits, Relevancy Search, Natural Language Processing, Probabilistic Machine Learning