About

Arpan Somani is an experienced AI professional currently working as an Applied Scientist - II at Amazon Alexa. He holds an M.Tech degree in Computer Science and Engineering from the prestigious Indian Institute of Technology, Bombay, with expertise in Data Science, Machine Learning, and NLP. During his masters, he presented his research work at top international conferences, namely NAACL and EMNLP. Arpan is skilled in C++, Python, Tensorflow, Pytorch, sklearn, Classical ML algorithms (Support Vector Machine, Decision Trees, Regression), Deep Learning algorithms (Convolutional neural network, Recurrent neural network, Deep Neural Networks). His tech stack includes Research Scientist, AI, NLP, ML, Neural Networks, Deep Learning, NLU, SVM, Python, Scikit, Data Scientist, Tensorflow, Search, Pytorch, Java, Elastic Search. Before joining Amazon, Arpan worked as a Research Engineer at Amex AI labs, where he worked on Text Mining and NLP problems as part of the Unstructured data team. He developed and deployed an ML-based pipeline for record linkage and entity-matching, a use case of credit fraud detection, which resulted in an annual saving of $6Mn. He also implemented LSH/Elastic Search-based text matching frameworks for entity resolution and prospect merchant identification. Arpan experimented and developed metrics to measure data quality from short noisy text from customer call transcripts used in downstream tasks. Arpan has also worked as an Application Engineer at Oracle-Cloud supply chain management development group. He envisaged NLP-based semantic search methods to improve upon keyword-based search for Oracle's B2B purchasing app. He resolved critical issues, optimized SQLs, and modularized code blocks, which helped in better real-time performance of the application. Overall, Arpan Somani has 6.61 years of relevant experience and a strong background in AI, NLP, and ML. He has a proven track record of developing and deploying ML-based pipelines and frameworks that have resulted in significant cost savings for organizations.