Jul 2021 – Jan 2024 · 2 yrs 6 mos · Bangalore Urban, Karnataka, India

• CPG Credit Risk Modelling -
• XX Mn USD reduction in bad debt beverage loans by identifying risky
• customers in Middle America Zone.
• The solution is deployed on Azure Cloud following MLOps principles
• Tech Stack - Python, Pyspark, Ensemble Modelling techniques, Databricks, Azure Machine Learning (Model Productization)
• Experimentation Platform -
• Build a product to automate experiment design, management, and uplift
• analysis between treatment and control groups. Improved the ease of rolling out a pilot with the reduced turnaround time to a few hours.
• Tech Stack - Python, Pyspark, Azure Pipelines, Databricks, Counterfactuals, A/B Testing, Azure Databricks Integration
• Backtesting Platform -
• Build a product to automate the backtesting of multiple versions of the model, to
• choose the best model for the pilot. Reduced turnaround time from weeks to a few hours
• Tech Stack - Python, Pyspark, PowerBI, Databricks

Jan 2020 – Sep 2021 · 1 yr 8 mos

• Reduced shipping cost by XX $ by predicting the number of containers to be booked with logistic vendor and helping the International supply chain team to avoid paying penalty for spot booking of the containers.
• The solution is built by stacking multiple algorithms (Arima, Xgboost, Lightgbm) and deployed on AWS cloud and consumed by business connecting through endpoint URL
• Annual savings of XX$ by identifying the market size for the new design launched in Apparels
• The solution is built by stacking multiple algorithms (CNN, Xgboost, Lightgbm) and deployed on AWS cloud and consumed by business connecting through endpoint URL
• Email personalization recommendation engine - XX % improvement in conversion rate post implementing the research paper-based (Triple2Vec) NLP algorithm and solution is integrated to Salesforce to roll out personalized emails to customers
• Deep Neural Call scanner [POC] - Idea selected in top 40 out of 800 Ideas in an international level hackathon organized by Telstra - Australia

Dec 2018 – Jan 2020 · 1 yr 1 mo

Jul 2017 – Dec 2018 · 1 yr 5 mos · Bengaluru Area, India

• Developed an ML lifecycle framework using AWS services/Dockers and MLFlow to keep track of ML experimentation, packaging a DS code in such a format to reproduce runs on any platform with easy deployment.
• Retail Pack Size optimization to cut down labor handling and inventory holding costs.
• Tech Stack used: AWS Cloud Services, Python, Pyspark, Docker-Containers, Pyomo, MLFlow

Apr 2016 – Jul 2017 · 1 yr 3 mos · Mumbai Area, India

• Experience in Data mining foundation and Data Visualization tools i.e. R, Python,SPSS, Sisense, MS Excel, SQL, and Tableau.
• Developed Text Mining algorithm to tag the videos having similar content using Python
• Developed NLP algorithm that provides in-depth analysis for the YouTube and Facebook comments using Python
• Experience in developing in-depth analytical reports for individual YouTube channels to provide insights on content performance, industry and competitor actionable recommendations that drive business decisions using R and Python
• Experience in understanding and manipulating raw data captured (using SQL/R) from various digital media platforms to generate automated reports (in Excel/Sisense) which is incorporated into the media intelligence platform
• Develop extensive reports and scalable R models to measure top KPIs showcasing category, YouTube channel and video performance across time for real-time feedback on content published which helps the Sales, BD, and Marketing team to drive the business decisions.
• Worked with internal functions such as Sales, Business Development etc. to develop R tools which will propagate self-serve for their business intelligence needs.
• Applied Knowledge of setting up AdWords Campaign as well as measuring and optimizing performance of campaign.

Jul 2012 – Jul 2015 · 3 yrs · Pune Area, India

• Experience in predictive modelling using Data mining techniques i.e. Linear and Logistic regression, Decision tree analysis, Bagging/Boosting, and Association analysis ( Marketing mix modelling)
• Led a BO Module Client Interaction to resolve issues, BO Module output and stabilised. Created a mapping of which issue should go to which team resulting in a 20 Hours saving per issue raised and aggregate streamline issue resolution
• Saved 3100 Man hours in a year, by implementing an Automation Scripting project. This project was recognised by the client at the all hands

Sep 2011 – Apr 2012 · 7 mos