Oct 2021 – Dec 2023 · 2 yrs 2 mos

• Worked on L1 & L2 list page ranking models, Ad ranking models, product list page personalisation & LLM use cases at Myntra.
• Worked on an ML model for the L2 Ranking Layer, aimed at real time re-ranking top products based on user preferences. The model incorporates user & product embeddings, in-session features, and dynamic product features, optimising for CTR. Resulted in significant uplift in business metrics
• Created a machine learning model for user-attribute recommendation, utilising a custom implementation of Bayesian Personalized Ranking (BPR) in Tensorflow with a weighted loss function and improved negative sampling techniques. Led to ~8% nDCG uplift. Also explored alternative models like LightFM, NCF & DeepFM.
• Runner-up prize in the Myntra Hackathon - Built an MVP for personalised banner creatives on Myntra's homepage.

Jul 2019 – Sep 2021 · 2 yrs 2 mos

• Designed and implemented a location-based recommendation system utilizing BPR model to learn location embeddings, which are subsequently employed in multiple downstream tasks (L1 ranking model, Trending near you). Led to significant uplift in revenue.
• Conducted experiments with frameworks like Nvidia's Triton and ONNX for scaling machine learning model serving and deployment.
• Implemented an improved diversity layer optimising for maximal marginal relevance (MMR) in the ranking pipeline.

May 2018 – Jul 2018 · 2 mos · Bengaluru Area, India

• Performed data-preprocessing & analytical techniques to retrieve useful features from large SQL database and then used standard ML techniques for clustering & building a prediction model.