Mar 2021 – Oct 2021 · 7 mos

• Drove long-term investment decisions (in tune of $2 million) for a new product by forecasting its revenue for a period of 30 months using Prophet library
• Calculated the value of this new product in improving the conversion rate
• Researched and developed a template for designing and conducting switchback testing enabling Course Hero to conduct unbiased A/B tests in the marketplace for any use case
• Won the Course Hero hackathon in the best product idea category by demonstrating a working solution for a business problem that impacts 20% of all subscribers

Jan 2020 – Mar 2021 · 1 yr 2 mos

• Led teams on several projects in the healthcare and life sciences verticals
• Improved drug sale forecasting model and reduced error to 1.46%, reducing the error by 55%
• Analyzed 200 million claims and identified fraudulent claims worth $100 million and 500 highest risk fraudulent actors using ML

Jun 2019 – Jan 2020 · 7 mos

• Interfaced with multiple stakeholders - including client operations manager, IT team - to get a consensus on the problem statement
• Successfully led a 4-member team to develop a machine learning based supply chain planning tool Proof-of-Concept for a multi-billion dollar pharmaceutical client
• The tool identifies patterns in demand, providing weekly recommendations to modify supply and manufacturing to match the demand
• The supply chain tool predicted the demand accurately, avoided 4 stockout situations due to spikes in demand and reduced warehouse inventory level by 40%
• Incedo's PoC was picked for a full scale project, out of multiple competitors

Sep 2018 – Jun 2019 · 9 mos · Washington D.C. Metro Area

• Developed a machine learning model to identify fraudulent insurances applications in Federally Facilitated Marketplace
• Identified cases of ID theft using a Naive Bayes text classifier to parse the transcript of a call with a customer representative

Oct 2016 – Sep 2018 · 1 yr 11 mos

• Structural Optimization using Evolutionary Algorithms (ongoing research)
• Used optimization heuristics to optimize structural designs obtained from topology optimization
• Developed EC toolbox in Python for optimization of structural designs
• Designed operators to derive new stable structural designs from existing ones. Designed algorithm to ensure stability of the derived structure
• Improved design space exploration by injecting creative structural designs
• Obtained optima improved prior solutions by 1.4\% and 61\% for the test cases
• Topology Optimization for Structural Design
• TO has been used as computational approach for component design. Combined TO with novel image processing based algorithm to design structures for the first time via TO
• Designed structures using TO, performance was comparable with solutions from literature
• Combined graph algorithms with model fitting to identify joints (98% accuracy) and structural members (100% accuracy) in TO structure
• Key contribution: Linked computational component design methods with structural optimization
• Published journal paper and conference paper based on this algorithm

Aug 2014 – Oct 2016 · 2 yrs 2 mos

• Bridge Type Classification
• Cleaned and repurposed National Bridge Inventory (NBI) data to predict design type of bridges across the United States
• Integrated NBI data with USGS seismic intensity dataset and historic cost data for concrete and steel to improve classifier performance
• Performed classification using Decision Tree and Bayesian Network algorithm and achieved precision and recall of over 88%
• Journal paper describing published in the Journal of Computing in Civil Engineering; Conference paper published in the proceedings of International Conference on Computing in Civil and Building Engineering 2016
• Optimizing the NFL Season Schedule
• Modeled the NFL match scheduling problem using integer programming in Gurobi
• Implemented branch and bound algorithm to maximize schedule popularity
• Developed metrics to numerically measure game's popularity
• Reduced time to optimality from 7 to 3 hrs by efficient constraint modeling
• Obtained schedule was judged second best out of 30 schedules

Dec 2012 – Jul 2014 · 1 yr 7 mos · Mumbai Area, India

• Co-founded and led Team Shunya, the first Indian team selected among 20 teams in SDE, to design and construct a life-size zero-energy 700 sq. ft. solar house
• Conceived and executed construction schedule to build house in 10 days and disassemble it in 4 days
• Led night shift (12 students) to construct house at competition; devised strategies on the fly to minimize construction delays due to parts damaged in shipping
• Awarded Special Mention for sustainability; 2nd rank among first time participants