Experience

Aakraya Research
• India
• Capital Markets
• 1 - 100 Employee
• Quantitative Researcher

  • Dec 2020 - Present

Auquan
• United Kingdom
• Financial Services
• 1 - 100 Employee
• Data Scientist

  • Jul 2019 - Dec 2020


Goldman Sachs

• United States
• Financial Services
• 700 & Above Employee

• ML Strat Analyst

  • Jul 2018 - Jul 2019

  Worked in Core ML team under Prof. Charles Elkan Worked in Core ML team under Prof. Charles Elkan



Indian Institute of Technology, Delhi

• Higher Education
• 700 & Above Employee

• Graduate Teaching Assistant

  • Aug 2017 - May 2018

  Working as head TA for the Computer Vision Course. Responsible for creating assignments, grading tests and answering queries of 75+ students Working as head TA for the Computer Vision Course. Responsible for creating assignments, grading tests and answering queries of 75+ students

• Summer Strat Analyst

  • May 2017 - Jul 2017

  Team : • Interned with People Analytics team in the Technology Division, Bangalore • The team worked towards using machine learning models to help the firm more effective policies for employees Work Profile : • Extensive data analytics, feature engineering, validating models, machine learning • FullStack development of data pipeline for automation of analysis and effective visualization • Technologies: R, Slang, Angular Details : • Carried out a comprehensive literature survey to identify factors that would help in predicting the performance of employees at individual and team level • Defined the concept of team at the optimal granularity level of 5-10 members using the employee data, used for measuring the overall team health and collaboration level among the members • Used employee review data for calculating the engagement of employees within and outside the team, identified and correlated the importance of both with the team and individual performance • Built a relationship graph over the firm to identify the significantly important employees and studied the effect of their voluntary termination on the attrition in their relationship circle • Studied the effect of gender and regional diversity on the performance of the team • Analyzed the factors such as location, proximity from leadership, academic background, direct or lateral hires etc on the performance of employees and their team • Created a data pipeline to digest data and carry out the analysis for any snapshot of time • Developed a browser-based Organization visualization tool for the senior leadership to view the hierarchical Org structure at any point in time. Huge improvement over the previous tool in loading time (from seconds to milliseconds) • Offered PPO (Pre-Placement Offer) by the firm for work done during the course of internship Show less



Microsoft

• United States
• Software Development
• 200 - 300 Employee

• Software Engineering Intern

  • May 2016 - Jul 2016

  Team : • Interned with Bing Local Team, Bing STCI at Microsoft IDC, Hyderabad • The team worked towards making information about local entities more easily accessible • The work revolved around identifying the query intent correctly and showing relevant results Work Profile: • Improving the quality of Local results shown by the Bing for the German market • Machine learning and NLP with main focus on feature engineering and data cleaning Details : • Worked towards improving the performance of local category query classifier for the German market. This involved identifying local category queries such as 'Hotels in Berlin', 'Restaurants in Berne' from the set of all queries • Created a full fledged Context free grammar to identify to required patterns like 'Hotels/restaurants in some place' or 'hospitals/fire stations near me' • Identified features like query length, presence of location, whether business name present etc • Compiled a comprehensive list of business names to be used as a negative signal and passed it through extensive data cleaning for reducing the number of false negatives • Created a tool for breaking the long German words into their constituents for better identification and reducing the complexity of the model • Trained a MART model over the new features and also using the signals from other classifiers, identified the significant features and further reduced model complexity • Precision of new model 14% more than the previous model, with 0% loss in recall • Explored the possibility of completely revamping the classifier by testing different models like word2vec and LSTM frameworks Show less