About

Aditya Gupta is an Applied Scientist at Flipkart with almost 4 years of practical experience in Machine Learning and Deep Learning. He has a major in Electrical Engineering from the prestigious Indian Institute of Technology, Delhi. Aditya has a strong background in Data Science, Machine Learning, Deep Learning, NLP, and Computer Vision, and is proficient in PyTorch and Tensorflow 2. He is also an active contributor to GitHub, where he shares his projects and code. At Flipkart, Aditya is currently working on building a Competitive Intelligence Platform (CIP) to measure the performance of Flipkart in comparison to other key players like Amazon, Bigbasket, Nykaa, etc., across different Business Units (BU), based on data extracted from four major social media platforms - Twitter, YouTube, Facebook, Instagram. Aditya's expertise in NLP has been instrumental in identifying a pre-defined set of customer experience nodes (L1s) with their sentiment, clustering the actual pain points of customers in each L1 into a dynamic list (static for now) of L2s using an unsupervised algorithm, and tagging each datapoint with a business unit (a pre-defined set of BUs), if applicable, based on the product mentioned in the text. These tags are used to generate scores at the Aspect x BU x Organisation level. Aditya's proficiency in Classical Machine Learning has helped him use CIP scores from the last step to predict the NPS of next week with MAPE of 1.2% and Market Share of the current month for all the BUs (8 for now) with MAE of 1.5% using an ensemble of Linear regression models. Prior to Flipkart, Aditya worked as an ML Software Engineer at Samsung Research Institute, where he worked on a project aimed at recognizing characters, emojis, hand gestures, etc., that one could create by moving fingers in the air without wearing any sensors. Aditya's contributions to this project included training a MobileNet version of DenseASPP from scratch on four different datasets to do the hand segmentation with an accuracy of 98%, using the curriculum learning strategy, and designing an encoder-decoder architecture using a novel combination of 2D & 3D Dense convolution to do a real-time hand gesture detection with a validation accuracy of 90%. Aditya's education background includes a Bachelor of Technology in Electrical Engineering from the Indian Institute of Technology, Delhi.