About

Sourabh Singhal is an accomplished Data Scientist with over 2 years of relevant experience. He completed his Integrated Dual degree from IIT Kharagpur and has a passion for problem-solving through coding. Sourabh is well-versed in Python, C, and has a flair for Mathematics and Statistics. He is currently employed at Sprinklr as a Data Scientist. Sourabh has a proven track record of delivering high-quality work. As a Machine Learning Intern at Quantiphi, he developed a generic pipeline architecture to process EHR data, capable of giving predictions like disease reoccurrence or 30-day survival. He built the input data processing pipeline using Apache Beam Python SDK, and pre-trained TensorFlowHub's Elmo embeddings were used. Sourabh also developed a parameterized LSTM model using high-level TensorFlow APIs to enable scalability and distributed training on GCP's AI platform. He achieved one-push deployment of the pipeline by automating steps from preprocessing to prediction using state-of-the-art GCP services. During his time at Vinod Gupta School of Management, Sourabh worked on a term project on stock price prediction using Hidden Markov Model. He implemented HMM Baum-welch and EM hill climbing algorithms to find unknown parameters and used the forward-backward algorithm to predict the future state of stock price movement. Sourabh's model predicted near future price movements in SBI stock with about 80% accuracy for the year 2016. Sourabh's education history includes a Dual Degree - BTech+Mtech in Manufacturing Science from IIT Kharagpur. He is also Level 1 qualified from CFA Institute and has completed a course from the Institute of Actuaries of India. Sourabh's tech stack includes Data Scientist, Research Scientist, Python, TensorFlow, Scikit, Keras, GCP, Computer Vision, ML, Pytorch, NLP, AI, Big Data, Neural Networks, SQL, BERT, Sentiment Analysis, Hyperparameter, Deep Learning, Javascript, Pandas, Web, and Senior Software Engineer. With his expertise in these areas, Sourabh is well-equipped to tackle complex data problems and deliver effective solutions.