May 2024 – Jun 2025 · 1 yr 1 mo · Bengaluru, Karnataka, India

May 2022 – May 2024 · 2 yrs · Bengaluru, Karnataka, India

May 2020 – May 2022 · 2 yrs · Bengaluru, Karnataka, India

Jul 2017 – May 2020 · 2 yrs 10 mos · Bengaluru, Karnataka, India

• Working on end to end ML workflows including data analysis and POC, data extraction, data preparation, hyper parameter tuning, feature engineering, model training pipeline and scalable model deployment for various recommendation features.
• Trending feed: Used XGBoost based Learning-To-Rank model (LambdaMART) to build this new feature.
• Built a meta data based item-item similarity model for news clips using a DNN in Tensorflow.
• Built sub-models to devise sampling strategy and feature engineering for different content types
• Solved Multi Arm Bandit problem using UCB for exploration of cold start news clips.
• Watch Next (Autoplay): Built LSTM based sequential model for movies, entertainment clips and TV shows.
• Masthead: Used Random Forest based sub model to engineer sports content specific features used in main ensemble model.
• `You May Also Like` trays: Developed CF and content filtering based hybrid item-item recommendation algorithms. Setup and productionized the initial Machine Learning pipeline in Apache Spark from scratch to combine multiple recommendation algorithms into a single workflow.

Jun 2017 – Aug 2017 · 2 mos · London, United Kingdom

• Worked on predictive modeling of remittance prices and speed between
• different countries based on various economic, geographic and demographic factors.

May 2017 – Jul 2017 · 2 mos · Singapore

• Worked on Object Recognition using transfer learning and Image segmentation using the concepts of Deep Learning.

May 2016 – Jul 2016 · 2 mos · Bengaluru Area, India

• I was on the Enterprise Computing Team at the Bangalore office of Morgan Stanley and worked on the development of Inter-Switch Link Monitor spanning all the switches in Storage Area Networks available with the firm globally.
• Project pipeline was as follows:
• Used SQL queries embedded in Perl to extract the traffic data, over a week, of all the Inter switch links from the production database.
• Organized the data in java objects and performed the calculations of threshold breaches in terms of percentile as required by the team.
• The final data needed on portal was ported through REST services using HTTP.
• Used AngularJS to develop the final portal including all the necessary features as required.

May 2015 – Jul 2015 · 2 mos · New Delhi

• My internship revolved around working on the Computer Vision Algorithms and open source libraries of C++. Project was reconstruction of 3D point cloud from 2D images / videos using SFM and SIFT algorithms and concluding an input acquisition by testing the code on different datasets.